AI transcript
0:00:03 Welcome to the Next Wave Podcast.
0:00:06 I’m Matt Wolfe, and today I’m joined once again by Matthew Berman,
0:00:12 one of the sharpest minds covering AI and the creator behind Forward Future Live,
0:00:16 a show that’s become an absolute must-watch for anyone trying to make sense
0:00:18 of where artificial intelligence is really headed.
0:00:23 Matt’s been on the front lines at major tech events like Dreamforce and BoxWorks,
0:00:25 talking directly with the people building the future.
0:00:29 In this episode, we’re going to dig into some of the biggest developments
0:00:30 from the past week.
0:00:34 Everything from Adobe’s new AI tools and what they mean for Creative Pros
0:00:38 to NVIDIA’s massive expansion and its surprising political moves
0:00:44 to OpenAI’s bold new timeline for creating a self-improving AI researcher.
0:00:47 We also dive into one of the most viral tech stories of the year,
0:00:51 the Neo-Humanoid Robot, and whether we’re actually ready to have one of them
0:00:53 in our own homes or not.
0:00:57 It’s a fun, fast-moving conversation packed with insights, hot takes,
0:01:01 and a few wild predictions about where AI and robotics are headed next.
0:01:05 So let’s just jump on over and chat with Matthew Berman.
0:01:11 Being a know-it-all used to be considered a bad thing,
0:01:14 but in business, it’s everything.
0:01:18 Because right now, most businesses only use 20% of their data,
0:01:22 unless you have HubSpot, where data that’s buried in emails,
0:01:27 call logs, and meeting notes become insights that help you grow your business.
0:01:29 Because when you know more, you grow more.
0:01:32 Visit HubSpot.com to learn more.
0:01:40 Hey, Matt, welcome back to the show.
0:01:41 Great to have you again.
0:01:42 Yeah, thanks, Matt.
0:01:45 I am very grateful to be here again.
0:01:47 I always like talking about AI with you.
0:01:48 Yeah, absolutely.
0:01:52 And you and I, we just had lunch a couple days ago up in San Francisco.
0:01:54 I was up at TechCrunch Disrupt,
0:01:58 and you generously stepped away from your office and production schedule
0:02:00 to come hang out and grab lunch with me.
0:02:04 So that was awesome to hang out and talk shop and YouTube strategy
0:02:06 and stuff like that for a good hour and a half or so.
0:02:08 Absolutely, yeah.
0:02:11 And it’s fun when people who watch our videos come up to us as well,
0:02:13 and they’re like, oh, you’re a YouTuber.
0:02:14 Oh, you’re a YouTuber.
0:02:15 Oh, I like your channel.
0:02:16 It’s like, oh, thanks, man.
0:02:18 We kind of had that experience at lunch where somebody came up
0:02:21 and said something to one of us
0:02:23 and then turned to the other one and went, you’re here too.
0:02:26 You know, like, so that was pretty cool.
0:02:26 That was pretty cool.
0:02:30 So in my opinion, this week was sort of a slower week,
0:02:31 but before we hit record, I was like,
0:02:33 what are the things you’re interested in?
0:02:34 And you rattled off like seven things.
0:02:36 So I’m like, oh, well, maybe there’s more happening this week
0:02:37 than I realized.
0:02:40 But saying that, I was at TechCrunch Disrupt,
0:02:42 and so I thought it would be fun to kind of dive in
0:02:44 and get caught up on the news this week
0:02:47 and just sort of chat and share our thoughts on it.
0:02:47 Yeah, absolutely.
0:02:50 There was really four sort of events this week
0:02:53 that were kind of going on in the AI sphere, right?
0:02:55 You had Adobe Max, which was going on in LA.
0:02:59 You had NVIDIA GTC, Washington, D.C.
0:03:02 I didn’t realize they were doing a second GTC this year
0:03:03 because they did one back in March,
0:03:06 but apparently they did one in Washington, D.C.
0:03:09 My sort of tinfoil hat theory around that
0:03:11 was it was for very political reasons,
0:03:12 but we can get into that if you want.
0:03:15 And of course, Jensen did another two-hour keynote.
0:03:16 Yeah.
0:03:18 There was also TechCrunch Disrupt,
0:03:20 which we’ll talk about briefly,
0:03:22 but there’s like no news that came out of it.
0:03:24 And then also OpenAI did a live session
0:03:27 that was sort of like a surprise session this week
0:03:29 where they talked about their nonprofit
0:03:32 and some of their timelines and all of that kind of stuff.
0:03:35 So those are some of the avenues I want to go down.
0:03:36 There was also the Neo robot
0:03:39 that would be fun to talk about a little bit.
0:03:41 So I said there wasn’t a lot to talk about,
0:03:45 but as I recap here, there’s quite a bit to talk about.
0:03:50 Yeah, and speaking of Neo, over 50 million views on X alone.
0:03:51 So something has resonated.
0:03:53 I am excited to talk about that with you.
0:03:54 Yeah, for sure.
0:03:56 Well, I’ll just kind of go through the list real quick.
0:03:58 So Adobe Max happened this week.
0:04:01 I don’t know how checked in you were with Adobe Max.
0:04:03 I know you tend to probably lean a little bit more
0:04:08 towards the LLMs versus like the creative side of things.
0:04:09 Yeah, fill me in.
0:04:10 What happened at Adobe Max?
0:04:13 I didn’t really feel like it was as big of updates as this year,
0:04:17 but they announced like a conversational AI assistant, right?
0:04:20 So like you’ll be able to go into Photoshop pretty soon
0:04:22 and just tell it, edit this image.
0:04:24 So the guy is wearing a banana on his head
0:04:27 or edit this image so that the background is the beach
0:04:30 instead of the desert or whatever, right?
0:04:33 So they’re adding in this like conversational chat assistant.
0:04:36 I think they saw like the success of Nano Banana
0:04:38 and a lot of these like AI editing tools
0:04:41 and then went, let’s just build that into our tools
0:04:43 and sort of like get ahead of the narrative.
0:04:45 I feel like that was kind of the main thing.
0:04:47 They also announced that they’re going to be doing
0:04:51 some like custom fine tuning for enterprises and for companies.
0:04:55 So if you want like a Firefly model trained on your branding,
0:04:58 your imagery, like your color scheme, that sort of stuff,
0:05:01 they will fine tune a Firefly image model for you.
0:05:03 There was a handful of other smaller announcements,
0:05:05 but I think those were kind of like the two
0:05:07 sort of highlight announcements for me.
0:05:08 Yeah.
0:05:09 I mean, let’s talk about the elephant in the room.
0:05:12 It does seem like there’s an existential threat
0:05:14 to Adobe’s business, right?
0:05:14 Right.
0:05:16 I mean, we have Nano Banana.
0:05:22 We have, you know, the plethora of other image generation models
0:05:24 and it’s so easy to use.
0:05:26 But then we have Photoshop,
0:05:28 which, you know, you have to actually learn the tool.
0:05:30 You can’t just speak in natural language.
0:05:34 And so I totally understand they’re rolling out features
0:05:39 that allow anybody to use kind of what they’ve been exposed to lately
0:05:41 with these models to actually apply that technique
0:05:46 of just prompting to some kind of amazing generation within Photoshop.
0:05:47 It makes a ton of sense.
0:05:50 It’s interesting because they really are serving
0:05:52 two separate audiences with that.
0:05:55 I suspect for image models,
0:05:57 we’re not having professional designers
0:05:59 and professional Photoshoppers, whatever.
0:06:03 They’re not also using these AI models.
0:06:03 Right.
0:06:06 And so if they’re building all of this functionality into Photoshop,
0:06:08 into their other products,
0:06:12 they have two potentially conflicting sets of audiences that they’re serving.
0:06:13 They have the professionals
0:06:16 and then they have the amateurs and just the tinkerers.
0:06:17 Right.
0:06:19 And it’s going to be interesting to see how this plays out.
0:06:22 I am a huge fan of focus in general.
0:06:28 And it just seems like you’re going to have this divergence of products
0:06:31 where it’s serving two different types of users
0:06:34 and it’s maybe not going to be the best at either of them.
0:06:34 Totally.
0:06:37 I totally see that take and I agree with it.
0:06:41 I feel like Photoshop should probably stay that suite of tools
0:06:44 for like the professionals that want to use those tools.
0:06:47 I will probably get in trouble from Adobe for saying this,
0:06:52 but like I’m really, really not a fan of Adobe’s business model either.
0:06:52 Right.
0:06:56 Like they’ve got the monthly billing and then like there’s actual fees to cancel it.
0:07:00 Now saying that you do agree to that when you sign up.
0:07:02 So everybody that bitches about the fact that like,
0:07:03 oh, I just I’m paying a monthly fee
0:07:06 and then they’re charging me $50 to cancel the monthly fee.
0:07:09 Well, when you sign up, they give you two options, right?
0:07:12 They’re like, you know, pay this amount just to get it for the year
0:07:16 with no recurring or pay this amount to essentially rent it from us.
0:07:17 And there is an early termination fee.
0:07:20 People always take the cheaper price and then get pissed off
0:07:22 that they have to pay the early termination fee.
0:07:26 So I do feel like Adobe does tell people and warn people.
0:07:27 And then like a year later,
0:07:29 people get pissed off that there’s an early termination fee.
0:07:34 But saying that it just creates this really crappy narrative around Adobe,
0:07:36 where everybody just thinks Adobe is this like evil company.
0:07:40 That’s like trying to like extort you for as much money as possible.
0:07:43 And I don’t really like that business model.
0:07:44 The other thing I was going to say about Adobe
0:07:46 that will probably get me in trouble with Adobe is
0:07:47 Yeah, keep it coming.
0:07:49 We did an episode here on the next wave
0:07:51 where we did like our tier list,
0:07:53 where we went like, you know, S tier all the way down
0:07:55 to like F tier of like image models.
0:07:55 Oh, no.
0:07:57 Out of all the models we talked about,
0:08:00 Adobe Firefly was the only one that fell into F tier.
0:08:01 Oh, man.
0:08:02 Yeah.
0:08:03 So I think there’s a few things there,
0:08:05 but I want to ask you first,
0:08:09 do you think this new crop of generative image models
0:08:15 is going to eat the lunch of Adobe, Adobe Photoshop specifically?
0:08:20 I really do personally think that there’s two different sets of users here,
0:08:24 but there’s totally a world in which I think
0:08:27 the AI image models of the world could get good enough
0:08:31 where you’re essentially just able to prompt anything that you want.
0:08:32 But what do you think?
0:08:36 Do you think that’s going to be severe competition for them in the future?
0:08:37 Or is it just two different sets?
0:08:39 Yeah, I don’t really know yet
0:08:42 because it does feel like whenever this stuff comes out,
0:08:44 Adobe is trying to stay ahead of it and go,
0:08:45 let’s just bake it into our software.
0:08:49 But I feel like we’re going to quickly get to a point where like,
0:08:52 well, why do I need to pay for Adobe to get that feature?
0:08:55 If I can just do it, you know, directly on the seed dream website
0:08:59 or directly on the, you know, Google AI studio with nano banana.
0:09:02 Like, why do I even need Adobe Photoshop anymore?
0:09:03 I don’t know.
0:09:06 I do think there’s going to still always be that subset of people
0:09:10 that are like, I love really getting into like the granular details
0:09:14 and like dialing in this composition that I really like that I have in my mind.
0:09:19 But I also feel like Adobe is sort of like isolating those people
0:09:24 because those are the types of people that tend to be more on the anti-AI side, right?
0:09:27 Like the people that really like to use Photoshop for what it’s designed for
0:09:31 and get granular, those are the people that are like, I hate AI.
0:09:34 And if Adobe is going to roll AI into Photoshop, right?
0:09:37 I’m going to use a tool that’s not going to use AI.
0:09:41 And so I feel like the core customers that they have right now
0:09:43 are going to start to rebel against Adobe.
0:09:48 And then the sort of casual users of Adobe that just want the like nano banana features.
0:09:51 Well, pretty soon, Adobe is not even going to be necessary for them
0:09:55 because these tools are going to get good enough where I can just like highlight the boots
0:09:58 and say, change them to red boots or whatever, you know?
0:09:58 Yeah.
0:10:03 So my last point on this is I actually think, and I’ve been thinking about this a long time,
0:10:07 the delineation between these different types of products, whether it’s image generation
0:10:10 or Photoshop more for professionals, they’re going to start to get blurred
0:10:13 and they’re going to start to look a lot alike.
0:10:21 And it’s going to enable a huge new crop of users and tinkers to really make incredible things.
0:10:25 But ultimately, and this is what I’ve really realized lately,
0:10:30 taste and the ability to curate an amazing human experience,
0:10:34 whether that is the design or an actual physical experience,
0:10:39 that’s going to matter more than anything and so much more going forward.
0:10:42 Of course, like the first thing that comes to mind is the Rick Rubin meme, right?
0:10:49 So like the ability to really know your taste and know who it serves and actually have good taste
0:10:55 in a world where anybody can create incredible visuals is going to be so important.
0:10:56 Yeah.
0:10:57 No, I totally agree with that.
0:10:59 I don’t know what Adobe is doing.
0:11:04 You know, I invest in a lot of stocks of companies that are making big moves in AI.
0:11:07 Adobe is not one of the companies I’ve put money into.
0:11:08 I’ll just say that.
0:11:13 I do have concerns about where Adobe is going to go in the future and,
0:11:15 you know, sort of losing their moat.
0:11:20 I do think Photoshop itself, maybe like Photoshop, maybe even Adobe Illustrator,
0:11:23 some of these kinds of tools.
0:11:26 Those, I think, are going to get replaced by AI pretty quickly.
0:11:29 Some of the tools like Adobe Premiere, Adobe After Effects,
0:11:35 probably going to be around a little bit longer because, you know, video is lagging a little
0:11:36 bit behind, you know, image editing.
0:11:41 Like we can’t take one of our YouTube videos that we ramble for, let’s say, an hour, give
0:11:45 it to an AI and say, cut this down to a 20 minute video that people actually want to watch,
0:11:45 you know?
0:11:46 Oh my God.
0:11:47 That’s my dream, Matt.
0:11:49 Where is that product?
0:11:56 I’ve seen hints of it here and there, but no, we’re still manually editing every single
0:11:57 video down to the frame.
0:12:03 And yeah, like even if we had 10, 20% help from AI, that would be amazing.
0:12:04 But no, we have not seen that yet.
0:12:04 Yeah.
0:12:09 I know Descript claims that that’s like what they’re working on, but the glimpses I’ve seen
0:12:11 from Descript are still pretty far off.
0:12:11 It’s hard.
0:12:14 It’s just the video as a modality is just hard to work with though.
0:12:16 Yeah, for sure.
0:12:18 Well, I’m going to shift over to TechCrunch Disrupt.
0:12:20 That was where I was the past few days.
0:12:26 So my perspective on TechCrunch Disrupt, like what it was, was essentially a whole bunch
0:12:27 of like fairly new startups.
0:12:32 There was a couple big companies represented, but a majority of them were like really, really
0:12:37 fresh startups, like within their first year or two, still working on like seed rounds,
0:12:39 like very, very early raises.
0:12:43 They all set up like little booths around the expo floor.
0:12:47 And as people walk around, they pitch why you should be interested in their company.
0:12:50 It feels like most of them are there because they want to raise capital.
0:12:54 So they’re just kind of putting out as many feelers as they can and talking to as many people
0:12:54 as they can.
0:12:59 Ideally, if somebody likes what they’re putting out there, they’ll get some investment.
0:13:05 And then they also do like this startup battle where founders get up on stage and sort of pitch
0:13:06 their idea.
0:13:09 And then it gets narrowed down, narrowed down, narrowed down until they pick their
0:13:13 one finalist, probably most famously seen in Silicon Valley.
0:13:19 When I’m so happy you brought that up at, oh, that was like peak TechCrunch Disrupt, peak
0:13:21 kind of Silicon Valley culture.
0:13:26 And for those of you watching who haven’t seen that show, Silicon Valley highly recommended.
0:13:27 Yeah.
0:13:28 Still a good watch.
0:13:31 It is a good watch and pretty accurate to the event.
0:13:37 I would say that Silicon Valley made out TechCrunch Disrupt to be a much bigger event than it actually
0:13:38 is in real life.
0:13:39 I think it used to be.
0:13:41 But other than that, pretty accurate.
0:13:41 Yeah.
0:13:45 It may have been more culturally relevant back then.
0:13:47 It’s slightly less now.
0:13:47 Yeah.
0:13:50 When I was there, I was fairly underwhelmed.
0:13:53 Like it was a lot smaller of an event than I expected.
0:13:58 When you go to San Francisco, you’ve got the Moscone Convention Center, which is essentially
0:14:00 broken out into like three buildings, right?
0:14:03 Well, they only used one of the buildings for TechCrunch Disrupt.
0:14:06 And in my mind, I was thinking they were going to be using all three.
0:14:10 Like when I went out there for GDC, the Game Developers Conference, that stretched all of
0:14:12 the convention center buildings, like the whole thing.
0:14:15 And I was kind of anticipating that with TechCrunch.
0:14:20 But I mean, TechCrunch Disrupt was basically like one of the three floors of Moscone, right?
0:14:21 Yeah.
0:14:24 And I don’t think we’re the only ones thinking it was underwhelming and maybe not worth the
0:14:25 time.
0:14:27 I don’t know if you want to show Gary Tan’s tweet.
0:14:28 This was Gary’s tweet.
0:14:30 TechCrunch Disrupt is a terrible waste of time.
0:14:33 And I don’t understand why any of you would go to it.
0:14:34 Well, I’ll tell you why I was there.
0:14:35 I was sponsored.
0:14:41 JetBrains sponsored me to come out and film their session that they gave and then also film
0:14:44 around their booth and make some content around JetBrains.
0:14:50 So I was there specifically for JetBrains, but had I not been paid to be there, would
0:14:51 I go back next year?
0:14:53 Probably not.
0:14:53 Yeah.
0:14:54 Yeah.
0:14:57 There’s a legacy to it.
0:15:03 I think one little coder down below as a reply is like, that’s pretty prescient.
0:15:08 TechCrunch after Michael Arrington, the founder of TechCrunch, sold is generally PR Newswire,
0:15:11 just a way to get your news out there and nothing more than that.
0:15:14 But I don’t know, there’s a history to TechCrunch Disrupt.
0:15:17 It makes me a little sad that it is not as big as it once was.
0:15:18 Yeah.
0:15:22 In my mind, I was almost thinking it was going to be like a mini CES kind of thing where companies
0:15:26 would actually use it as an opportunity to make announcements and things like that.
0:15:27 It really wasn’t.
0:15:31 And if any company was making announcements, it was all just like very, very small announcements.
0:15:36 Like we added this new feature to our API where you can now grab this data that you
0:15:38 didn’t use to be able to grab.
0:15:41 And people are like, whoo, you know, like nothing too exciting.
0:15:42 Yeah.
0:15:43 Yeah.
0:15:51 Hey, if you take a look at my web presence online, it’s safe to say that I’m a bit AI
0:15:51 obsessed.
0:15:55 I even have a podcast all about AI that you’re watching right now.
0:16:00 I’ve gone down multiple rabbit holes with AI and done countless hours of research on the
0:16:03 newest AI tools every single week.
0:16:04 Well, I’ve done it again.
0:16:07 And I just dropped my list of my favorite AI tools.
0:16:11 I’ve done all the research on what’s been working for me, my favorite use cases and more.
0:16:16 So if you want to steal my favorite tools and use them for yourself, now you can.
0:16:18 You can get it at the link in the description below.
0:16:19 Now back to the show.
0:16:28 So the other event I wanted to chat about quickly was we had NVIDIA GTC in Washington, D.C.
0:16:30 this week as well.
0:16:34 It was a tough week because I was invited to three out of these four events, Adobe, TechCrunch,
0:16:37 and NVIDIA and ended up going to TechCrunch.
0:16:43 And probably from the videos and footage I saw from the other two would have selected to go
0:16:44 to one of the other two had I had the option.
0:16:48 But during the NVIDIA GTC, you’ve probably seen more of the keynote than I have.
0:16:51 I actually have not had an opportunity to watch it yet.
0:16:56 But the big things for me that came out of it were the new IGX Thor, which is like the
0:16:59 brain that they’re trying to put into like robots and things like that.
0:17:02 And then also, this is something I don’t totally understand.
0:17:07 You might know a little bit more about it, but it seemed powerful just in his explanation
0:17:14 was the open source Arial software, which is going to create this like AI native 6G wireless
0:17:14 network.
0:17:16 It’s a little bit over my head.
0:17:17 I’ll be totally honest.
0:17:20 But those were the two things that sort of grabbed my attention from his keynote.
0:17:22 But I’m curious what you saw in the keynote.
0:17:25 So I only watched some of it.
0:17:31 I think the news that stood out to me was they’re taking a huge stake in Nokia.
0:17:34 First of all, OK, I didn’t know Nokia was still around.
0:17:34 That’s cool.
0:17:40 You know, I definitely had one of those Nokia bricks that lasted years and played Snake on.
0:17:43 And yeah, so like I used to be a huge fan of Nokia.
0:17:48 And then the iPhone basically demolished every other phone maker that was around at the time.
0:17:50 So that was surprising.
0:17:52 They invested a billion dollars, if I remember correctly.
0:17:55 They got a chunk of stock from them.
0:18:02 And it seems like NVIDIA is making these investments into so many different companies.
0:18:09 And it goes back to this like kind of roundabout profit and revenue cycle that we’re seeing.
0:18:15 So I think another good example is NVIDIA invested a hundred billion into OpenAI.
0:18:21 And then OpenAI is going to take that money and essentially spend it on NVIDIA chips.
0:18:27 And so I’m like, all of this counts towards, you know, like the country’s economy.
0:18:30 But like, aren’t we double counting dollars at this point?
0:18:33 And so they’ve done a number of these types of investments.
0:18:39 There have been a lot of memes, a lot of graphs showing the very incestuous nature of these top
0:18:45 tech companies, whether it’s Frontier Labs or hardware creators like AMD and NVIDIA.
0:18:48 But I just thought, so bring it back to Nokia.
0:18:50 It’s interesting to see what they’re going to do there.
0:18:56 They did say, speaking of 6G, that that’s something they’re working on with Nokia, if I remember correctly.
0:18:57 Interesting.
0:18:58 Interesting.
0:19:02 I wonder, so is Nokia still making phones?
0:19:06 Is that still their model or have they moved on to something else?
0:19:08 Are they like making phone chips?
0:19:12 Like, I’m not actually quite sure what Nokia is doing because I don’t think I’ve seen a Nokia
0:19:15 phone on the market anytime recently.
0:19:17 I’m not actually sure what they do anymore.
0:19:18 I haven’t seen a Nokia phone.
0:19:21 You’re either buying an Android or an iPhone.
0:19:22 Obviously, iPhone’s Apple.
0:19:28 But Android, you see like Huawei, you see Samsung, you see Google, but not much else.
0:19:29 Yeah, yeah, yeah.
0:19:34 There’s a few like smaller outliers like the nothing phone and there’s a few others.
0:19:38 But yeah, it’s mostly Google, Samsung and Apple are like the players you see.
0:19:40 I’m looking it up right now.
0:19:41 I’m trying to find if they’re still.
0:19:47 So according to ChadGPT, yes, the Nokia brand still appears on phones, but no, Nokia Corporation
0:19:49 doesn’t directly manufacture them.
0:19:54 They sold their mobile phone business to Microsoft in 2014.
0:19:59 A finished company, HMD Global holds the license for Nokia branded phones.
0:20:02 And so, no, it doesn’t really seem like they’re making it anymore.
0:20:03 So Nokia is owned by Microsoft?
0:20:06 Their mobile phone business is Microsoft.
0:20:09 I don’t even think it’s like branded Nokia anymore, but I’m not sure.
0:20:10 Ah, okay.
0:20:11 That’ll be really interesting.
0:20:16 Yeah, that’s something that’s definitely fascinated me about like NVIDIA is NVIDIA’s strategy
0:20:22 of investing in these various companies, but then that investment money turning around and
0:20:23 essentially buying NVIDIA chips.
0:20:24 Yeah.
0:20:26 And they just passed a five trillion, right?
0:20:28 They passed five trillion dollars in market cap.
0:20:29 Yeah.
0:20:29 Yeah, yeah.
0:20:32 So, I mean, that’s absolutely wild to see.
0:20:34 And I’m going to be honest.
0:20:40 I know there’s talk about an AI bubble and there probably is an AI bubble, but I feel like NVIDIA
0:20:42 as one of the companies is going to make it through.
0:20:44 Yeah, for sure.
0:20:46 Yeah, there’s been a lot of talk of AI bubble lately.
0:20:52 And the way I see it, and obviously, look, I’ve dedicated my working life to AI at this point.
0:20:52 Right.
0:20:59 The potential value of artificial intelligence seems nearly infinite to me.
0:21:01 It just seems like there’s so much that can be done.
0:21:11 So even if we’re over-investing today in infrastructure, in models, if most companies have bigger inference
0:21:16 bills than what they’re actually charging customers, I think all this math is going to work out in
0:21:17 the end.
0:21:23 And, of course, there’s an analogy to be drawn in kind of the early and mid-2010s where, you know,
0:21:30 companies like Uber, Airbnb, Lyft, these companies were just burning through VC capital, subsidizing
0:21:33 rides, subsidizing housing, and they made it, right?
0:21:36 And there’s a thousand examples of those that didn’t.
0:21:45 But the VC math is typically over-invest, capture the market, and then monetize more based on business
0:21:46 fundamentals later on.
0:21:47 Yeah, yeah, yeah.
0:21:51 I mean, Amazon, didn’t Amazon lose money for like 20 years or something like that?
0:21:51 Exactly.
0:21:55 And all they had to do was turn the dial, and all of a sudden, they had margins.
0:21:56 So it’s going to be interesting.
0:22:00 Maybe we’re in a mini bubble, but in the long run, I just don’t think we’re going to see this
0:22:03 huge drop like a dot-com bubble.
0:22:08 And there’s been a number of interesting tweets about this, where if you look at the price of
0:22:11 earnings, we’re nowhere near the dot-com era.
0:22:12 Yeah, yeah, yeah.
0:22:15 You were showing me that graph, I think, last Friday when we were doing a live stream together.
0:22:18 It’s really, really interesting to me.
0:22:22 Did you hear that, like, NVIDIA’s now got a stake in Intel, which is something that I never
0:22:23 saw coming either.
0:22:25 No, I didn’t see that.
0:22:30 I know the U.S. government has a stake in Intel, which is kind of nuts.
0:22:30 Yeah.
0:22:33 Well, that’s probably why NVIDIA went in, let’s be honest.
0:22:37 They probably saw the government got invested in Intel, and NVIDIA’s like, all right, we’re
0:22:37 in as well.
0:22:42 And, you know, I mentioned that GTC, the second version of GTC this year was in Washington,
0:22:43 D.C.
0:22:46 Well, I think they did that for, like, political reasons.
0:22:49 I think he did that to be close to Trump.
0:22:54 Obviously, Jensen, from what I understand during his keynote, was very, you know, favorable towards
0:22:56 Trump and how Trump is handling things.
0:22:57 Well, of course.
0:23:00 He seemed to be fairly okay with the whole China situation.
0:23:01 Yeah.
0:23:06 Look, I think Zuckerberg is probably the most prominent example of this, of just doing a
0:23:07 180.
0:23:11 These guys who lead the Mag 7, they are incredibly bright.
0:23:15 You know, ultimately, they’ll do whatever it takes to make the company successful.
0:23:18 And if that means kissing the ring of the current administration or whatever the next administration
0:23:20 is, they will do that.
0:23:23 And whether they believe it or not doesn’t really matter.
0:23:24 They’re going to do it.
0:23:29 And that is how they’re currying favor with the administration.
0:23:30 Yeah.
0:23:35 No, it’s been kind of funny, a little slightly sad to watch, but, you know, interesting nonetheless.
0:23:36 Expected.
0:23:37 Yeah, yeah.
0:23:42 I mean, seeing, like, Tim Cook hand that, like, clear disk thing to Trump.
0:23:47 And I think Jensen handed Trump, like, a signed, I don’t know, like, one of his signed GPUs
0:23:48 or something.
0:23:52 I don’t remember what he gave him, but he gave him something to get in his favor.
0:23:55 But yeah, no, it’s really interesting to see how this is all playing out, because I would
0:23:59 have thought that Jensen would really be fighting for the business in China, but he seems to be
0:24:01 okay just sort of letting that go.
0:24:02 Yeah.
0:24:05 Well, I assume he’s still going to be fighting.
0:24:09 I think China’s making it difficult to sell into China, though, because they see the writing
0:24:10 on the wall.
0:24:16 They know that if they build their AI infrastructure on top of U.S. chips and software, they are
0:24:20 going to be at a disadvantage and beholden to the U.S.
0:24:22 And of course, they don’t want that.
0:24:22 Yeah, yeah.
0:24:24 Totally makes sense.
0:24:29 Okay, let me tell you about another podcast I know you’re going to love.
0:24:34 It’s called Billion Dollar Moves, hosted by Sarah Chin Spellings.
0:24:38 It’s brought to you by the HubSpot Podcast Network, the audio destination for business
0:24:39 professionals.
0:24:45 Join venture capitalist and strategist Sarah Chin Spellings as she asks the hard questions and
0:24:49 learns through the triumphs, failures, and hard lessons of the creme de la creme, so you
0:24:53 too can make billion dollar moves in venture, in business, and in life.
0:24:59 She just did a great episode called The Purpose Driven Power Player Behind Silicon Valley’s
0:25:01 Quiet Money with Mike Anders.
0:25:04 Listen to Billion Dollar Moves wherever you get your podcasts.
0:25:11 Well, let’s talk quickly, too, about the OpenAI event that happened this week.
0:25:15 So they did sort of like an, I don’t know, what would you call the live stream?
0:25:17 A sort of state of OpenAI kind of update thing?
0:25:19 It felt like ad hoc.
0:25:19 I don’t know.
0:25:23 I mean, obviously, they had been planning it a while, but it just felt like out of nowhere.
0:25:27 And then, yeah, there was kind of a few minutes about the, like an overview of it.
0:25:28 It was kind of very technical.
0:25:31 Jakob Pachocki was there with Sam Altman.
0:25:34 Jakob being the chief scientist, I believe is his title.
0:25:35 And then they did a Q&A.
0:25:36 Yeah.
0:25:39 So Sam went to X and posted this TLDR.
0:25:42 I’m not going to read the whole thing because even his TLDR could use a TLDR.
0:25:47 Honestly, I think the first line, we have set internal goals.
0:25:51 If you want to read that, that is crazy important to me.
0:25:52 All right, let’s see.
0:25:58 So we have set internal goals of having an automated AI research intern by September of
0:26:04 2026, running on hundreds of thousands of GPUs and a true automated AI researcher by March
0:26:06 of 2028.
0:26:10 We may totally fail at this goal, but given the extraordinary potential impacts, we think it
0:26:13 is in the public interest to be transparent about this.
0:26:20 So in the talk of AGI, it sounds like Sam Altman has the belief that AGI is coming before
0:26:28 2028, but an automated AI researcher by March of 2028, like if AGI comes sooner than that,
0:26:30 wouldn’t the AI researcher also come sooner than that?
0:26:31 You know what I mean?
0:26:32 Okay.
0:26:36 There’s so much to unpack from this one paragraph here.
0:26:39 So first, he does not mention AGI.
0:26:41 And I actually think that’s important to your point.
0:26:46 They have ironed out a new relationship, a new dynamic, a new deal with Microsoft.
0:26:54 And a big part of the original deal with Microsoft was Microsoft had ownership of OpenAI’s IP up
0:26:56 until they hit AGI.
0:26:58 Now, AGI was such an amorphous term.
0:27:00 It was a moving goalpost.
0:27:04 Nobody truly knows what AGI means or how to measure it.
0:27:06 So kind of ignored that.
0:27:08 So there’s two other things I want to point out here.
0:27:14 One, these are incredibly specific dates for saying that they’re going to have something
0:27:18 like an AI researcher, an automated AI researcher.
0:27:19 September, 2026.
0:27:20 Okay.
0:27:24 That’s, you know, not too far away, but it is a very precise date.
0:27:30 The one that blew my mind is March, 2028 to have a true automated AI researcher.
0:27:34 It is a specific month within a specific year.
0:27:35 It just seems wild.
0:27:42 Like, what does Sam Altman know to have such a precise prediction of that timeline?
0:27:50 Now, the last thing that I want to point out is once a company has a true automated AI researcher,
0:27:57 he didn’t say any of this, but I’m going to infer that also means you’re going to have self-improving
0:27:58 artificial intelligence.
0:28:04 And once you have self-improving artificial intelligence, the only limiter for really just
0:28:11 an explosion of intelligence to use the situational awareness paper is how much compute can you throw
0:28:12 at it?
0:28:12 Right?
0:28:18 And so it seems to me, if you have these, you know, handful of companies racing towards
0:28:25 whatever that goal is, the self-improving AI seems to be really the only goal that matters
0:28:30 because once one of the companies hits that goal, they are forever in the lead.
0:28:35 And it just seems very obvious to me that I know Zuckerberg is heading towards that.
0:28:37 He calls it artificial super intelligence.
0:28:42 Sam Altman is calling it an automated AI researcher, but they’re all talking about the same thing.
0:28:46 They’re talking about this point at which AI can improve itself.
0:28:46 Right.
0:28:47 Right.
0:28:52 And to me, that’s also the point where, you know, AI has the potential to be the, you know,
0:28:54 quote unquote, runaway AI.
0:28:54 Right.
0:28:59 That’s the point where AI, if it’s self-improving, it might start to figure out like that’s where
0:29:04 the sort of paperclip maximizer theoretical experiments come into play is okay.
0:29:06 Well, I’m sort of optimizing for this goal.
0:29:08 So what do I have to do to achieve the goal?
0:29:11 You know, possibly at the expense of everything else.
0:29:13 That’s where things get scary.
0:29:16 The reason I bring up AGI is I know Sam Altman did an interview last year.
0:29:19 Maybe he was just being hype Altman.
0:29:19 Right.
0:29:23 But in an interview last year, when somebody said, what are you excited for in 2025?
0:29:25 He said, I’m excited about AGI in 2025.
0:29:26 Right.
0:29:27 Something along those lines.
0:29:29 He said, I’m excited about AGI next year.
0:29:33 So he almost implied last year that he thought we’d have AGI this year.
0:29:39 And now he’s saying, you know, the automated AI researcher by March of 2028, which seems like
0:29:42 if you have AGI before March of 2028.
0:29:46 Are those the same level, AGI and self-improving AI?
0:29:50 Like, would you see those as sort of like being the same thing or are those two different milestones?
0:29:55 So Matt, they seem, and based on my understanding, they are two completely different milestones.
0:29:58 Maybe not completely different, but they are very different.
0:30:03 And so when he says AGI, if we’re using his and open AI’s kind of previous definition of AGI,
0:30:12 it is essentially when AI can accomplish most economically valuable work better than the average or most humans.
0:30:15 I butchered it a little bit, but I think that was the definition.
0:30:16 And I actually do think we’re almost there.
0:30:19 Granted, we have two more months left in this year, so we’ll see.
0:30:25 But I think he got to a point with the Microsoft relationship where he couldn’t really use or
0:30:27 like have a hard definition of AGI.
0:30:34 Now, going towards self-improving artificial intelligence, a true AI researcher, that seems
0:30:35 very different.
0:30:44 That is when AI can discover novel algorithms and new ways to build silicon and then apply
0:30:46 it to itself, it becomes more powerful.
0:30:52 And thus, the next piece of research just happens more quickly or becomes, you know, a more important
0:30:53 discovery.
0:30:57 So I do think those are two very different things.
0:31:02 And AGI, in the context of his previous definition, does seem like it is right around the corner.
0:31:07 And in fact, a lot of these models, the core intelligence of the models, I think we’re there.
0:31:12 I just think we’re now needing to build the scaffolding and the implementation around it,
0:31:14 which seems to be lagging by about 18 months.
0:31:17 So the core intelligence of the model is there for AGI.
0:31:18 We just need to go implement it now.
0:31:22 So I don’t think he was wrong with his predictions.
0:31:25 It’s just the definitions that he’s using are pretty amorphous.
0:31:27 Yeah, yeah, I totally agree with that.
0:31:31 Yeah, I remember you were at Dreamforce this year doing your forward future live during Dreamforce.
0:31:35 And one of the terms I kept on hearing you bring up over and over again was capability
0:31:36 overhang.
0:31:37 Yeah, exactly.
0:31:40 It sounds to me like that’s kind of what you’re referring to here, right?
0:31:44 It’s actually smarter than what most people are using it for right now, right?
0:31:45 Yeah, exactly.
0:31:47 The models are extremely capable.
0:31:55 The scaffolding around them, the tools, you know, memory tool use, access to different information
0:31:58 sources, all of that stuff still needs to continue to be built out.
0:32:04 Yeah, along the same lines here, I just pulled up this article, both OpenAI and Microsoft posted
0:32:08 this article on their website about the next chapter of Microsoft OpenAI partnership, which
0:32:14 ties into what we’re saying, because they basically said once AGI is declared by OpenAI, the declaration
0:32:17 will now be verified by an independent expert panel.
0:32:23 So they’re going to put together some sort of panel to decide, OK, did OpenAI actually achieve
0:32:23 AGI.
0:32:28 So, I mean, it’s still subjective, but it’s now left up to the subjectivity of this panel
0:32:32 as opposed to like Sam Altman just claiming we got AGI.
0:32:33 Interesting.
0:32:35 So this just came out, right?
0:32:36 A couple of days ago.
0:32:36 Yeah.
0:32:37 Oh, interesting.
0:32:39 So maybe I got that wrong.
0:32:44 Actually, I thought the whole notion of you get the IP.
0:32:48 Microsoft OpenAI says you get the IP until we declare AGI.
0:32:52 But it seems like they still do have that kind of going up to this paragraph.
0:32:53 I can read real quick.
0:32:57 The agreement preserves key elements that have fueled this successful partnership, meaning
0:33:00 OpenAI remains Microsoft’s frontier model partner.
0:33:08 OK, Microsoft continues to have exclusive IP rights and Azure API exclusivity until AGI.
0:33:10 I might have made a mistake there.
0:33:16 I thought they had removed that part of the agreement, but it seems like they just clarified
0:33:22 that OpenAI alone doesn’t get to declare it, although they do get to declare when it happens.
0:33:25 They declare it and then an independent panel verifies it.
0:33:26 Right.
0:33:28 And so let’s look at the incentives here.
0:33:35 OpenAI is incentivized to declare AGI as soon as possible because they do not want to give
0:33:38 Microsoft exclusive IP rights, right?
0:33:40 They don’t want Azure to have exclusivity.
0:33:43 Any exclusivity is bad generally for a company.
0:33:49 And so OpenAI is definitely incentivized to claim AGI, and they could have done so at any
0:33:52 moment, but now they actually have to have a third party validate it.
0:33:52 Right.
0:33:54 And what that validation is based on, I don’t know.
0:33:55 Who knows?
0:33:55 Yeah.
0:34:01 But it does say here, Microsoft’s IP rights for both models and products are extended through
0:34:07 2032 and now include models post-AGI with appropriate safety guardrails.
0:34:13 So basically, it sounds like their agreement goes through 2032 regardless of AGI.
0:34:19 I’m immediately thinking of the Megamind meme and Satya Nadella just playing 4D chess.
0:34:21 This is what comes to mind.
0:34:24 I wish I could just instantly pull that meme up and show it right now, but you’re just going
0:34:25 to have to visualize it.
0:34:26 Yeah.
0:34:30 I mean, it seems like Microsoft worked out a pretty nice deal for this.
0:34:31 They had all the leverage.
0:34:32 I’m not surprised.
0:34:36 Microsoft’s IP rights now exclude OpenAI’s consumer hardware.
0:34:41 So whatever they end up making with Johnny Ive, if they ever do get into robotics, that’s
0:34:43 going to be outside of Microsoft’s scope.
0:34:50 I do remember there was a little bit of tension around when OpenAI approached WindSurf to acquire
0:34:50 WindSurf.
0:34:54 It seemed like it was Microsoft that got that deal shut down, right?
0:34:56 Like Microsoft went, oh, you’re buying WindSurf?
0:35:01 Well, because of our agreement, that means we’ll also own a portion of WindSurf.
0:35:07 And OpenAI didn’t like that, but it does sound like there’s no exclusions for like acquired
0:35:10 software from OpenAI, just consumer hardware.
0:35:11 Yeah.
0:35:14 And again, like Microsoft just had all the leverage here.
0:35:19 They invested and got all of these terms before OpenAI was OpenAI, right?
0:35:19 Yeah.
0:35:20 I think it was 2017.
0:35:25 And by the way, did you see Satya recently did an interview in which he said, Bill Gates
0:35:30 thought we were just burning a billion dollars on that investment in OpenAI, which is like,
0:35:32 you know, no, I didn’t see that.
0:35:32 Oh my God.
0:35:40 Even the most prominent tech minds in our multiple generations get stuff wrong.
0:35:42 Nobody really knows what’s going on.
0:35:43 Maybe except for Satya.
0:35:46 Satya seems to be just operating on another level.
0:35:47 He does.
0:35:47 Yeah, yeah, yeah.
0:35:53 Also, Microsoft can now independently pursue AGI alone or in partnership with third parties.
0:35:58 So I mean, OpenAI and Microsoft are competitors now in the race to AGI.
0:35:59 Oh, absolutely.
0:36:02 They’re going to be competitors on almost every dimension, right?
0:36:03 OpenAI is building hardware.
0:36:09 It’s just like a completely new way to write and build software.
0:36:15 They can compete with every piece of Microsoft software all the way down to the operating system
0:36:15 level.
0:36:16 I truly believe this.
0:36:21 Microsoft sees OpenAI as an existential threat, but they’re also a partner.
0:36:24 They’re like, you know, keep your enemies close type thing.
0:36:25 Yeah, no, absolutely.
0:36:28 So let’s talk about the robots real quick here.
0:36:29 Let’s talk about robots.
0:36:34 We got the Neo home robot that was announced this week, which you can actually pre-order
0:36:35 for a $200 deposit.
0:36:37 Did you see this one at CES?
0:36:41 They actually had this one at CES and they were doing like a demo of it, like vacuuming
0:36:42 a living room kind of thing.
0:36:43 No, I think I missed that.
0:36:49 This one is just interesting because it’s like a robot wearing clothes to kind of hide
0:36:50 the fact that it’s a robot.
0:36:54 So do you want to quickly fill me in on what’s got you excited about the Neo robot?
0:36:55 Yeah, absolutely.
0:37:03 So this is really like the first time a company has officially opened orders for a robot that
0:37:08 you can buy and have in your home today, with the exception of Unitree, which is a Chinese
0:37:09 company.
0:37:12 I was going to say, I’ve seen like streamers and stuff that have like the Unitree robot.
0:37:15 Like, I don’t know if you saw the Kaisenat one where they were like pushing it around and
0:37:15 stuff.
0:37:16 It was awesome.
0:37:16 Yeah.
0:37:20 So yeah, it’s definitely possible to buy it, but you’re going to be buying it from China
0:37:22 and everything that comes with that.
0:37:28 So this is the first US company that you’re able to actually buy a humanoid robot that is
0:37:29 supposed to do chores around your house.
0:37:31 It is built for your house.
0:37:36 They show off a bunch of examples, cleaning, welcoming guests by opening the door, doing
0:37:41 laundry, cooking food, potentially getting medication for you.
0:37:45 I mean, there’s a thousand use cases and their launch of video, as I mentioned earlier, went
0:37:46 absolutely mega viral.
0:37:52 50 million plus views just on X alone, not including YouTube and Instagram and everywhere else they
0:37:53 probably launched.
0:38:00 Now, Joanna Stern, I believe her name is, actually had an in-person review and demo of this robot.
0:38:04 Turns out it is 100% tele-operated.
0:38:10 And what that means is there is somebody wearing VR goggles with handset of some sort operating
0:38:10 the robot.
0:38:15 So it is not yet true artificial intelligence inside the robot.
0:38:19 Now, that is not to take anything away from this accomplishment.
0:38:24 It is wild to think just a few years ago, this was a complete pipe dream.
0:38:29 And now we actually have working humanoid robots that seem to be quite impressive.
0:38:37 MKBHD just launched a video yesterday talking about how a lot of tech companies are now over-promising
0:38:38 and under-delivering.
0:38:40 This is kind of another prime example.
0:38:47 And then even like within tech, within AI, it’s even more prominent that companies are over-promising,
0:38:47 under-delivering.
0:38:50 Of course, he was talking about the rabbit and-
0:38:51 Humane.
0:38:52 A humane pin, right.
0:38:57 And so once again, the demos so far today showed that it is 100% tele-operated.
0:39:02 In the demo video, there were two different scenes in which it was fully autonomous.
0:39:03 That’s great.
0:39:07 But the promise is it is the majority of uses is going to be autonomous.
0:39:11 They said they’re going to be delivering this by the beginning of next year.
0:39:11 Yeah.
0:39:16 So it says right here, US deliveries begin in 2026, starting with early access customers.
0:39:18 It doesn’t say when in 2026.
0:39:21 It just says be in 2026.
0:39:24 So they basically have over a year to start delivering this.
0:39:26 I put in a pre-order, by the way.
0:39:26 I’m super excited.
0:39:28 I tend to be very optimistic.
0:39:31 I like thinking that this stuff is going to work.
0:39:33 But of course, they got the robot.
0:39:34 It works great.
0:39:35 It moves well.
0:39:38 But that last 10% to 20% always is the hard part.
0:39:41 And I think it’s very analogous to self-driving.
0:39:46 Elon and Tesla have been talking about delivering fully capable self-driving for over a decade.
0:39:53 And we’re now just kind of like starting to get to somewhere that it feels like, okay, he’s delivering on that promise.
0:39:57 But it was, you know, a year away at all for the last decade.
0:39:58 It seems really cool.
0:39:59 I’m into it.
0:40:08 The part that is maybe a little worrisome for me is if I have this humanoid robot in my house and it’s not able to do things autonomously.
0:40:13 There’s going to be somebody else operating this robot, seeing into my house, doing things in my house.
0:40:17 And that does not rub me the right way.
0:40:18 Interesting.
0:40:18 Okay.
0:40:20 I didn’t catch that when you said it the first time.
0:40:23 So I understood that they’re tele-operating it, but I assume that was for the demos.
0:40:29 But like if you get one of these and put it in your house, somebody is going to be tele-operating it while it cleans your house?
0:40:30 Yeah.
0:40:31 So it’s in the FAQs.
0:40:48 But basically what happens is if you give it a task and it is not able to accomplish it autonomously, which most likely is going to be a lot of tasks in the early days, you can schedule a time with their tele-operators for them to accomplish the task.
0:40:57 Now, the hope is by the tele-operator accomplishing the task through Neo, Neo will learn and not need a tele-operator for subsequent tasks.
0:41:08 But again, you are giving permission for this person running this robot that’s capable of lifting 150 pounds, by the way, to operate in your house, which is just, it’s wild.
0:41:14 I don’t know if I feel more uncomfortable with having an autonomous robot doing that or somebody operating it.
0:41:15 I think it’s the latter though.
0:41:16 Yeah.
0:41:24 I think somebody operating it is creepier to me that like somebody is looking around in my house, like know where I put my wallet, things like that.
0:41:24 Yeah.
0:41:25 Yeah.
0:41:32 But, you know, on the flip side of the coin, if you’re looking at like one silver lining, you know, everybody keeps on saying AI is going to create new jobs.
0:41:37 This is a new job that didn’t exist, tele-operating robots for people.
0:41:38 Yeah.
0:41:38 Yeah.
0:41:40 Matt, what do you think about the look?
0:41:43 What do you think about the design of the robot?
0:41:49 I think, honestly, it’s one of the less cool robots that I’ve seen.
0:41:51 What makes you say that?
0:41:51 I don’t know.
0:41:55 To me, it looks like a giant human-sized teddy bear or something.
0:42:06 Do you remember like a few years ago, Elon Musk went on stage to announce Optimus and they had a bunch of people with like Optimus outfits on dancing?
0:42:06 Yeah.
0:42:09 It feels like that, but there is actually a robot under it.
0:42:18 For me, like having the very soft look wrapped in cloth and just, it looks a little, I don’t know.
0:42:21 It’s not my design aesthetic that I would choose.
0:42:25 When I see figures robot, that is what I want robots to look like.
0:42:29 I want to see the metal and the different materials.
0:42:29 I agree.
0:42:30 That’s where I’m at too.
0:42:34 This to me looks like it’s from like a Tim Burton movie or something, if I’m being honest.
0:42:36 Oh, yeah, totally.
0:42:38 But anybody who’s wondering, I just looked up the price here.
0:42:44 So you can buy one for 20 grand or there’s a subscription for 500 bucks a month.
0:42:50 So like if you can’t afford the 20 grand up front, you can basically like rent one of these robots to live in your house.
0:42:54 So, Matt, let’s assume this is all going to work out.
0:42:56 It’s going to deliver on its promise.
0:42:58 Whether that happens or not, let’s make that assumption.
0:43:01 The 20K price point is very interesting.
0:43:08 That is essentially the lowest priced car in the United States is a little bit, I think it’s 18,000.
0:43:12 So it is essentially a very inexpensive car.
0:43:21 Plus, if you start layering on financing options, it really becomes something that I think a broad portion of the population can afford.
0:43:23 500 bucks a month.
0:43:35 OK, that seems like a lot of money, but imagine how much time you will save not having to clean up, not having to do the dishes, not having to tidy.
0:43:37 And I guess all of these use cases are really cleaning.
0:43:42 But I just think that the price point was the most compelling aspect of this launch.
0:43:52 Yeah, well, I mean, I figure if you hire like a housekeeper, right, like let’s say they come once a week, you’re probably going to pay 500 bucks a month for that housekeeper to come once a week.
0:43:56 This is someone that something that is there every day.
0:43:56 Right.
0:44:01 So if you were to hire a private chef, right, that’s a whole additional cost.
0:44:03 If you were to hire a housekeeper, that’s an additional cost.
0:44:13 Like if you were to hire out all of the individual things that this one robot theoretically will be able to do for you, this is actually the more economical option, it seems.
0:44:14 Did you preorder?
0:44:15 I haven’t yet, but I probably will.
0:44:17 I mean, it’s 200 bucks.
0:44:20 I tend to preorder pretty much every new tech gadget that comes out.
0:44:21 So I had a feeling.
0:44:22 Yeah, yeah.
0:44:23 I put in the preorder.
0:44:29 I’m going to go with the early access because that’s what I want and I want to show it off on camera.
0:44:40 So, uh, yeah, my Neo robot, after I learned the third world or piloting them is mad chill in video games with them in the house is a mess.
0:44:41 Excellent.
0:44:42 Just hanging out.
0:44:42 Yeah.
0:44:44 What do we got here?
0:44:46 Uh, day two of cleaning after humans.
0:44:47 Yeah.
0:44:49 My robot is going to see things.
0:44:50 Let’s just say that.
0:44:52 Yeah.
0:44:52 We’ve got kids.
0:44:54 There’ll be, uh, yeah.
0:44:58 Our robots are most likely to turn on humans because of what they have to see.
0:44:58 Yeah.
0:45:01 Uh, but yeah, this is hilarious.
0:45:05 I think this is a whole new category of meme and I’m ready for it.
0:45:06 Yeah, absolutely.
0:45:09 Well, Matt, this has been absolutely amazing.
0:45:13 I love having you on because you and I can nerd out about AI forever.
0:45:16 I feel like every time we have a conversation, it always ends the same way.
0:45:18 I feel like we could talk about this stuff forever.
0:45:18 Oh yeah.
0:45:20 We could easily keep going for another two hours.
0:45:21 Next time.
0:45:26 But, um, well, we’ll definitely have you back on if you’re willing to come back on and do this again.
0:45:27 Absolutely.
0:45:31 But it’s been a blast and, uh, I’m really loving what you’re doing with the forward future live.
0:45:31 Thank you.
0:45:35 I love those live streams that you’re doing and you’re rotating through guests and stuff.
0:45:35 Really, really, really.
0:45:35 Yeah.
0:45:37 We have to have you back on speaking.
0:45:37 Yeah.
0:45:38 It’s a blast.
0:45:39 It’s a blast.
0:45:39 So thank you.
0:45:40 Thanks so much.
0:45:41 Any place you want to shout out.
0:45:44 Obviously, everybody needs to check out Matthew Berman on YouTube.
0:45:45 Check out forward future.
0:45:48 That’s where he does the live show forward future live as well.
0:45:50 Anywhere else people should go after listening to this.
0:45:51 Yeah.
0:45:55 So if you can just click on the forward future Twitter thing right below Nick Wentz on the
0:45:59 right side of the screen, just follow us at forward underscore future underscore.
0:46:04 We’re working on getting a better handle, but for now it is that follow us at forward future.
0:46:04 Awesome.
0:46:05 Well, Matt, thanks so much.
0:46:07 And, uh, looking forward to doing it again with you in the future.
0:46:08 Thanks, Matt.
0:46:48 Thank you.
0:00:06 I’m Matt Wolfe, and today I’m joined once again by Matthew Berman,
0:00:12 one of the sharpest minds covering AI and the creator behind Forward Future Live,
0:00:16 a show that’s become an absolute must-watch for anyone trying to make sense
0:00:18 of where artificial intelligence is really headed.
0:00:23 Matt’s been on the front lines at major tech events like Dreamforce and BoxWorks,
0:00:25 talking directly with the people building the future.
0:00:29 In this episode, we’re going to dig into some of the biggest developments
0:00:30 from the past week.
0:00:34 Everything from Adobe’s new AI tools and what they mean for Creative Pros
0:00:38 to NVIDIA’s massive expansion and its surprising political moves
0:00:44 to OpenAI’s bold new timeline for creating a self-improving AI researcher.
0:00:47 We also dive into one of the most viral tech stories of the year,
0:00:51 the Neo-Humanoid Robot, and whether we’re actually ready to have one of them
0:00:53 in our own homes or not.
0:00:57 It’s a fun, fast-moving conversation packed with insights, hot takes,
0:01:01 and a few wild predictions about where AI and robotics are headed next.
0:01:05 So let’s just jump on over and chat with Matthew Berman.
0:01:11 Being a know-it-all used to be considered a bad thing,
0:01:14 but in business, it’s everything.
0:01:18 Because right now, most businesses only use 20% of their data,
0:01:22 unless you have HubSpot, where data that’s buried in emails,
0:01:27 call logs, and meeting notes become insights that help you grow your business.
0:01:29 Because when you know more, you grow more.
0:01:32 Visit HubSpot.com to learn more.
0:01:40 Hey, Matt, welcome back to the show.
0:01:41 Great to have you again.
0:01:42 Yeah, thanks, Matt.
0:01:45 I am very grateful to be here again.
0:01:47 I always like talking about AI with you.
0:01:48 Yeah, absolutely.
0:01:52 And you and I, we just had lunch a couple days ago up in San Francisco.
0:01:54 I was up at TechCrunch Disrupt,
0:01:58 and you generously stepped away from your office and production schedule
0:02:00 to come hang out and grab lunch with me.
0:02:04 So that was awesome to hang out and talk shop and YouTube strategy
0:02:06 and stuff like that for a good hour and a half or so.
0:02:08 Absolutely, yeah.
0:02:11 And it’s fun when people who watch our videos come up to us as well,
0:02:13 and they’re like, oh, you’re a YouTuber.
0:02:14 Oh, you’re a YouTuber.
0:02:15 Oh, I like your channel.
0:02:16 It’s like, oh, thanks, man.
0:02:18 We kind of had that experience at lunch where somebody came up
0:02:21 and said something to one of us
0:02:23 and then turned to the other one and went, you’re here too.
0:02:26 You know, like, so that was pretty cool.
0:02:26 That was pretty cool.
0:02:30 So in my opinion, this week was sort of a slower week,
0:02:31 but before we hit record, I was like,
0:02:33 what are the things you’re interested in?
0:02:34 And you rattled off like seven things.
0:02:36 So I’m like, oh, well, maybe there’s more happening this week
0:02:37 than I realized.
0:02:40 But saying that, I was at TechCrunch Disrupt,
0:02:42 and so I thought it would be fun to kind of dive in
0:02:44 and get caught up on the news this week
0:02:47 and just sort of chat and share our thoughts on it.
0:02:47 Yeah, absolutely.
0:02:50 There was really four sort of events this week
0:02:53 that were kind of going on in the AI sphere, right?
0:02:55 You had Adobe Max, which was going on in LA.
0:02:59 You had NVIDIA GTC, Washington, D.C.
0:03:02 I didn’t realize they were doing a second GTC this year
0:03:03 because they did one back in March,
0:03:06 but apparently they did one in Washington, D.C.
0:03:09 My sort of tinfoil hat theory around that
0:03:11 was it was for very political reasons,
0:03:12 but we can get into that if you want.
0:03:15 And of course, Jensen did another two-hour keynote.
0:03:16 Yeah.
0:03:18 There was also TechCrunch Disrupt,
0:03:20 which we’ll talk about briefly,
0:03:22 but there’s like no news that came out of it.
0:03:24 And then also OpenAI did a live session
0:03:27 that was sort of like a surprise session this week
0:03:29 where they talked about their nonprofit
0:03:32 and some of their timelines and all of that kind of stuff.
0:03:35 So those are some of the avenues I want to go down.
0:03:36 There was also the Neo robot
0:03:39 that would be fun to talk about a little bit.
0:03:41 So I said there wasn’t a lot to talk about,
0:03:45 but as I recap here, there’s quite a bit to talk about.
0:03:50 Yeah, and speaking of Neo, over 50 million views on X alone.
0:03:51 So something has resonated.
0:03:53 I am excited to talk about that with you.
0:03:54 Yeah, for sure.
0:03:56 Well, I’ll just kind of go through the list real quick.
0:03:58 So Adobe Max happened this week.
0:04:01 I don’t know how checked in you were with Adobe Max.
0:04:03 I know you tend to probably lean a little bit more
0:04:08 towards the LLMs versus like the creative side of things.
0:04:09 Yeah, fill me in.
0:04:10 What happened at Adobe Max?
0:04:13 I didn’t really feel like it was as big of updates as this year,
0:04:17 but they announced like a conversational AI assistant, right?
0:04:20 So like you’ll be able to go into Photoshop pretty soon
0:04:22 and just tell it, edit this image.
0:04:24 So the guy is wearing a banana on his head
0:04:27 or edit this image so that the background is the beach
0:04:30 instead of the desert or whatever, right?
0:04:33 So they’re adding in this like conversational chat assistant.
0:04:36 I think they saw like the success of Nano Banana
0:04:38 and a lot of these like AI editing tools
0:04:41 and then went, let’s just build that into our tools
0:04:43 and sort of like get ahead of the narrative.
0:04:45 I feel like that was kind of the main thing.
0:04:47 They also announced that they’re going to be doing
0:04:51 some like custom fine tuning for enterprises and for companies.
0:04:55 So if you want like a Firefly model trained on your branding,
0:04:58 your imagery, like your color scheme, that sort of stuff,
0:05:01 they will fine tune a Firefly image model for you.
0:05:03 There was a handful of other smaller announcements,
0:05:05 but I think those were kind of like the two
0:05:07 sort of highlight announcements for me.
0:05:08 Yeah.
0:05:09 I mean, let’s talk about the elephant in the room.
0:05:12 It does seem like there’s an existential threat
0:05:14 to Adobe’s business, right?
0:05:14 Right.
0:05:16 I mean, we have Nano Banana.
0:05:22 We have, you know, the plethora of other image generation models
0:05:24 and it’s so easy to use.
0:05:26 But then we have Photoshop,
0:05:28 which, you know, you have to actually learn the tool.
0:05:30 You can’t just speak in natural language.
0:05:34 And so I totally understand they’re rolling out features
0:05:39 that allow anybody to use kind of what they’ve been exposed to lately
0:05:41 with these models to actually apply that technique
0:05:46 of just prompting to some kind of amazing generation within Photoshop.
0:05:47 It makes a ton of sense.
0:05:50 It’s interesting because they really are serving
0:05:52 two separate audiences with that.
0:05:55 I suspect for image models,
0:05:57 we’re not having professional designers
0:05:59 and professional Photoshoppers, whatever.
0:06:03 They’re not also using these AI models.
0:06:03 Right.
0:06:06 And so if they’re building all of this functionality into Photoshop,
0:06:08 into their other products,
0:06:12 they have two potentially conflicting sets of audiences that they’re serving.
0:06:13 They have the professionals
0:06:16 and then they have the amateurs and just the tinkerers.
0:06:17 Right.
0:06:19 And it’s going to be interesting to see how this plays out.
0:06:22 I am a huge fan of focus in general.
0:06:28 And it just seems like you’re going to have this divergence of products
0:06:31 where it’s serving two different types of users
0:06:34 and it’s maybe not going to be the best at either of them.
0:06:34 Totally.
0:06:37 I totally see that take and I agree with it.
0:06:41 I feel like Photoshop should probably stay that suite of tools
0:06:44 for like the professionals that want to use those tools.
0:06:47 I will probably get in trouble from Adobe for saying this,
0:06:52 but like I’m really, really not a fan of Adobe’s business model either.
0:06:52 Right.
0:06:56 Like they’ve got the monthly billing and then like there’s actual fees to cancel it.
0:07:00 Now saying that you do agree to that when you sign up.
0:07:02 So everybody that bitches about the fact that like,
0:07:03 oh, I just I’m paying a monthly fee
0:07:06 and then they’re charging me $50 to cancel the monthly fee.
0:07:09 Well, when you sign up, they give you two options, right?
0:07:12 They’re like, you know, pay this amount just to get it for the year
0:07:16 with no recurring or pay this amount to essentially rent it from us.
0:07:17 And there is an early termination fee.
0:07:20 People always take the cheaper price and then get pissed off
0:07:22 that they have to pay the early termination fee.
0:07:26 So I do feel like Adobe does tell people and warn people.
0:07:27 And then like a year later,
0:07:29 people get pissed off that there’s an early termination fee.
0:07:34 But saying that it just creates this really crappy narrative around Adobe,
0:07:36 where everybody just thinks Adobe is this like evil company.
0:07:40 That’s like trying to like extort you for as much money as possible.
0:07:43 And I don’t really like that business model.
0:07:44 The other thing I was going to say about Adobe
0:07:46 that will probably get me in trouble with Adobe is
0:07:47 Yeah, keep it coming.
0:07:49 We did an episode here on the next wave
0:07:51 where we did like our tier list,
0:07:53 where we went like, you know, S tier all the way down
0:07:55 to like F tier of like image models.
0:07:55 Oh, no.
0:07:57 Out of all the models we talked about,
0:08:00 Adobe Firefly was the only one that fell into F tier.
0:08:01 Oh, man.
0:08:02 Yeah.
0:08:03 So I think there’s a few things there,
0:08:05 but I want to ask you first,
0:08:09 do you think this new crop of generative image models
0:08:15 is going to eat the lunch of Adobe, Adobe Photoshop specifically?
0:08:20 I really do personally think that there’s two different sets of users here,
0:08:24 but there’s totally a world in which I think
0:08:27 the AI image models of the world could get good enough
0:08:31 where you’re essentially just able to prompt anything that you want.
0:08:32 But what do you think?
0:08:36 Do you think that’s going to be severe competition for them in the future?
0:08:37 Or is it just two different sets?
0:08:39 Yeah, I don’t really know yet
0:08:42 because it does feel like whenever this stuff comes out,
0:08:44 Adobe is trying to stay ahead of it and go,
0:08:45 let’s just bake it into our software.
0:08:49 But I feel like we’re going to quickly get to a point where like,
0:08:52 well, why do I need to pay for Adobe to get that feature?
0:08:55 If I can just do it, you know, directly on the seed dream website
0:08:59 or directly on the, you know, Google AI studio with nano banana.
0:09:02 Like, why do I even need Adobe Photoshop anymore?
0:09:03 I don’t know.
0:09:06 I do think there’s going to still always be that subset of people
0:09:10 that are like, I love really getting into like the granular details
0:09:14 and like dialing in this composition that I really like that I have in my mind.
0:09:19 But I also feel like Adobe is sort of like isolating those people
0:09:24 because those are the types of people that tend to be more on the anti-AI side, right?
0:09:27 Like the people that really like to use Photoshop for what it’s designed for
0:09:31 and get granular, those are the people that are like, I hate AI.
0:09:34 And if Adobe is going to roll AI into Photoshop, right?
0:09:37 I’m going to use a tool that’s not going to use AI.
0:09:41 And so I feel like the core customers that they have right now
0:09:43 are going to start to rebel against Adobe.
0:09:48 And then the sort of casual users of Adobe that just want the like nano banana features.
0:09:51 Well, pretty soon, Adobe is not even going to be necessary for them
0:09:55 because these tools are going to get good enough where I can just like highlight the boots
0:09:58 and say, change them to red boots or whatever, you know?
0:09:58 Yeah.
0:10:03 So my last point on this is I actually think, and I’ve been thinking about this a long time,
0:10:07 the delineation between these different types of products, whether it’s image generation
0:10:10 or Photoshop more for professionals, they’re going to start to get blurred
0:10:13 and they’re going to start to look a lot alike.
0:10:21 And it’s going to enable a huge new crop of users and tinkers to really make incredible things.
0:10:25 But ultimately, and this is what I’ve really realized lately,
0:10:30 taste and the ability to curate an amazing human experience,
0:10:34 whether that is the design or an actual physical experience,
0:10:39 that’s going to matter more than anything and so much more going forward.
0:10:42 Of course, like the first thing that comes to mind is the Rick Rubin meme, right?
0:10:49 So like the ability to really know your taste and know who it serves and actually have good taste
0:10:55 in a world where anybody can create incredible visuals is going to be so important.
0:10:56 Yeah.
0:10:57 No, I totally agree with that.
0:10:59 I don’t know what Adobe is doing.
0:11:04 You know, I invest in a lot of stocks of companies that are making big moves in AI.
0:11:07 Adobe is not one of the companies I’ve put money into.
0:11:08 I’ll just say that.
0:11:13 I do have concerns about where Adobe is going to go in the future and,
0:11:15 you know, sort of losing their moat.
0:11:20 I do think Photoshop itself, maybe like Photoshop, maybe even Adobe Illustrator,
0:11:23 some of these kinds of tools.
0:11:26 Those, I think, are going to get replaced by AI pretty quickly.
0:11:29 Some of the tools like Adobe Premiere, Adobe After Effects,
0:11:35 probably going to be around a little bit longer because, you know, video is lagging a little
0:11:36 bit behind, you know, image editing.
0:11:41 Like we can’t take one of our YouTube videos that we ramble for, let’s say, an hour, give
0:11:45 it to an AI and say, cut this down to a 20 minute video that people actually want to watch,
0:11:45 you know?
0:11:46 Oh my God.
0:11:47 That’s my dream, Matt.
0:11:49 Where is that product?
0:11:56 I’ve seen hints of it here and there, but no, we’re still manually editing every single
0:11:57 video down to the frame.
0:12:03 And yeah, like even if we had 10, 20% help from AI, that would be amazing.
0:12:04 But no, we have not seen that yet.
0:12:04 Yeah.
0:12:09 I know Descript claims that that’s like what they’re working on, but the glimpses I’ve seen
0:12:11 from Descript are still pretty far off.
0:12:11 It’s hard.
0:12:14 It’s just the video as a modality is just hard to work with though.
0:12:16 Yeah, for sure.
0:12:18 Well, I’m going to shift over to TechCrunch Disrupt.
0:12:20 That was where I was the past few days.
0:12:26 So my perspective on TechCrunch Disrupt, like what it was, was essentially a whole bunch
0:12:27 of like fairly new startups.
0:12:32 There was a couple big companies represented, but a majority of them were like really, really
0:12:37 fresh startups, like within their first year or two, still working on like seed rounds,
0:12:39 like very, very early raises.
0:12:43 They all set up like little booths around the expo floor.
0:12:47 And as people walk around, they pitch why you should be interested in their company.
0:12:50 It feels like most of them are there because they want to raise capital.
0:12:54 So they’re just kind of putting out as many feelers as they can and talking to as many people
0:12:54 as they can.
0:12:59 Ideally, if somebody likes what they’re putting out there, they’ll get some investment.
0:13:05 And then they also do like this startup battle where founders get up on stage and sort of pitch
0:13:06 their idea.
0:13:09 And then it gets narrowed down, narrowed down, narrowed down until they pick their
0:13:13 one finalist, probably most famously seen in Silicon Valley.
0:13:19 When I’m so happy you brought that up at, oh, that was like peak TechCrunch Disrupt, peak
0:13:21 kind of Silicon Valley culture.
0:13:26 And for those of you watching who haven’t seen that show, Silicon Valley highly recommended.
0:13:27 Yeah.
0:13:28 Still a good watch.
0:13:31 It is a good watch and pretty accurate to the event.
0:13:37 I would say that Silicon Valley made out TechCrunch Disrupt to be a much bigger event than it actually
0:13:38 is in real life.
0:13:39 I think it used to be.
0:13:41 But other than that, pretty accurate.
0:13:41 Yeah.
0:13:45 It may have been more culturally relevant back then.
0:13:47 It’s slightly less now.
0:13:47 Yeah.
0:13:50 When I was there, I was fairly underwhelmed.
0:13:53 Like it was a lot smaller of an event than I expected.
0:13:58 When you go to San Francisco, you’ve got the Moscone Convention Center, which is essentially
0:14:00 broken out into like three buildings, right?
0:14:03 Well, they only used one of the buildings for TechCrunch Disrupt.
0:14:06 And in my mind, I was thinking they were going to be using all three.
0:14:10 Like when I went out there for GDC, the Game Developers Conference, that stretched all of
0:14:12 the convention center buildings, like the whole thing.
0:14:15 And I was kind of anticipating that with TechCrunch.
0:14:20 But I mean, TechCrunch Disrupt was basically like one of the three floors of Moscone, right?
0:14:21 Yeah.
0:14:24 And I don’t think we’re the only ones thinking it was underwhelming and maybe not worth the
0:14:25 time.
0:14:27 I don’t know if you want to show Gary Tan’s tweet.
0:14:28 This was Gary’s tweet.
0:14:30 TechCrunch Disrupt is a terrible waste of time.
0:14:33 And I don’t understand why any of you would go to it.
0:14:34 Well, I’ll tell you why I was there.
0:14:35 I was sponsored.
0:14:41 JetBrains sponsored me to come out and film their session that they gave and then also film
0:14:44 around their booth and make some content around JetBrains.
0:14:50 So I was there specifically for JetBrains, but had I not been paid to be there, would
0:14:51 I go back next year?
0:14:53 Probably not.
0:14:53 Yeah.
0:14:54 Yeah.
0:14:57 There’s a legacy to it.
0:15:03 I think one little coder down below as a reply is like, that’s pretty prescient.
0:15:08 TechCrunch after Michael Arrington, the founder of TechCrunch, sold is generally PR Newswire,
0:15:11 just a way to get your news out there and nothing more than that.
0:15:14 But I don’t know, there’s a history to TechCrunch Disrupt.
0:15:17 It makes me a little sad that it is not as big as it once was.
0:15:18 Yeah.
0:15:22 In my mind, I was almost thinking it was going to be like a mini CES kind of thing where companies
0:15:26 would actually use it as an opportunity to make announcements and things like that.
0:15:27 It really wasn’t.
0:15:31 And if any company was making announcements, it was all just like very, very small announcements.
0:15:36 Like we added this new feature to our API where you can now grab this data that you
0:15:38 didn’t use to be able to grab.
0:15:41 And people are like, whoo, you know, like nothing too exciting.
0:15:42 Yeah.
0:15:43 Yeah.
0:15:51 Hey, if you take a look at my web presence online, it’s safe to say that I’m a bit AI
0:15:51 obsessed.
0:15:55 I even have a podcast all about AI that you’re watching right now.
0:16:00 I’ve gone down multiple rabbit holes with AI and done countless hours of research on the
0:16:03 newest AI tools every single week.
0:16:04 Well, I’ve done it again.
0:16:07 And I just dropped my list of my favorite AI tools.
0:16:11 I’ve done all the research on what’s been working for me, my favorite use cases and more.
0:16:16 So if you want to steal my favorite tools and use them for yourself, now you can.
0:16:18 You can get it at the link in the description below.
0:16:19 Now back to the show.
0:16:28 So the other event I wanted to chat about quickly was we had NVIDIA GTC in Washington, D.C.
0:16:30 this week as well.
0:16:34 It was a tough week because I was invited to three out of these four events, Adobe, TechCrunch,
0:16:37 and NVIDIA and ended up going to TechCrunch.
0:16:43 And probably from the videos and footage I saw from the other two would have selected to go
0:16:44 to one of the other two had I had the option.
0:16:48 But during the NVIDIA GTC, you’ve probably seen more of the keynote than I have.
0:16:51 I actually have not had an opportunity to watch it yet.
0:16:56 But the big things for me that came out of it were the new IGX Thor, which is like the
0:16:59 brain that they’re trying to put into like robots and things like that.
0:17:02 And then also, this is something I don’t totally understand.
0:17:07 You might know a little bit more about it, but it seemed powerful just in his explanation
0:17:14 was the open source Arial software, which is going to create this like AI native 6G wireless
0:17:14 network.
0:17:16 It’s a little bit over my head.
0:17:17 I’ll be totally honest.
0:17:20 But those were the two things that sort of grabbed my attention from his keynote.
0:17:22 But I’m curious what you saw in the keynote.
0:17:25 So I only watched some of it.
0:17:31 I think the news that stood out to me was they’re taking a huge stake in Nokia.
0:17:34 First of all, OK, I didn’t know Nokia was still around.
0:17:34 That’s cool.
0:17:40 You know, I definitely had one of those Nokia bricks that lasted years and played Snake on.
0:17:43 And yeah, so like I used to be a huge fan of Nokia.
0:17:48 And then the iPhone basically demolished every other phone maker that was around at the time.
0:17:50 So that was surprising.
0:17:52 They invested a billion dollars, if I remember correctly.
0:17:55 They got a chunk of stock from them.
0:18:02 And it seems like NVIDIA is making these investments into so many different companies.
0:18:09 And it goes back to this like kind of roundabout profit and revenue cycle that we’re seeing.
0:18:15 So I think another good example is NVIDIA invested a hundred billion into OpenAI.
0:18:21 And then OpenAI is going to take that money and essentially spend it on NVIDIA chips.
0:18:27 And so I’m like, all of this counts towards, you know, like the country’s economy.
0:18:30 But like, aren’t we double counting dollars at this point?
0:18:33 And so they’ve done a number of these types of investments.
0:18:39 There have been a lot of memes, a lot of graphs showing the very incestuous nature of these top
0:18:45 tech companies, whether it’s Frontier Labs or hardware creators like AMD and NVIDIA.
0:18:48 But I just thought, so bring it back to Nokia.
0:18:50 It’s interesting to see what they’re going to do there.
0:18:56 They did say, speaking of 6G, that that’s something they’re working on with Nokia, if I remember correctly.
0:18:57 Interesting.
0:18:58 Interesting.
0:19:02 I wonder, so is Nokia still making phones?
0:19:06 Is that still their model or have they moved on to something else?
0:19:08 Are they like making phone chips?
0:19:12 Like, I’m not actually quite sure what Nokia is doing because I don’t think I’ve seen a Nokia
0:19:15 phone on the market anytime recently.
0:19:17 I’m not actually sure what they do anymore.
0:19:18 I haven’t seen a Nokia phone.
0:19:21 You’re either buying an Android or an iPhone.
0:19:22 Obviously, iPhone’s Apple.
0:19:28 But Android, you see like Huawei, you see Samsung, you see Google, but not much else.
0:19:29 Yeah, yeah, yeah.
0:19:34 There’s a few like smaller outliers like the nothing phone and there’s a few others.
0:19:38 But yeah, it’s mostly Google, Samsung and Apple are like the players you see.
0:19:40 I’m looking it up right now.
0:19:41 I’m trying to find if they’re still.
0:19:47 So according to ChadGPT, yes, the Nokia brand still appears on phones, but no, Nokia Corporation
0:19:49 doesn’t directly manufacture them.
0:19:54 They sold their mobile phone business to Microsoft in 2014.
0:19:59 A finished company, HMD Global holds the license for Nokia branded phones.
0:20:02 And so, no, it doesn’t really seem like they’re making it anymore.
0:20:03 So Nokia is owned by Microsoft?
0:20:06 Their mobile phone business is Microsoft.
0:20:09 I don’t even think it’s like branded Nokia anymore, but I’m not sure.
0:20:10 Ah, okay.
0:20:11 That’ll be really interesting.
0:20:16 Yeah, that’s something that’s definitely fascinated me about like NVIDIA is NVIDIA’s strategy
0:20:22 of investing in these various companies, but then that investment money turning around and
0:20:23 essentially buying NVIDIA chips.
0:20:24 Yeah.
0:20:26 And they just passed a five trillion, right?
0:20:28 They passed five trillion dollars in market cap.
0:20:29 Yeah.
0:20:29 Yeah, yeah.
0:20:32 So, I mean, that’s absolutely wild to see.
0:20:34 And I’m going to be honest.
0:20:40 I know there’s talk about an AI bubble and there probably is an AI bubble, but I feel like NVIDIA
0:20:42 as one of the companies is going to make it through.
0:20:44 Yeah, for sure.
0:20:46 Yeah, there’s been a lot of talk of AI bubble lately.
0:20:52 And the way I see it, and obviously, look, I’ve dedicated my working life to AI at this point.
0:20:52 Right.
0:20:59 The potential value of artificial intelligence seems nearly infinite to me.
0:21:01 It just seems like there’s so much that can be done.
0:21:11 So even if we’re over-investing today in infrastructure, in models, if most companies have bigger inference
0:21:16 bills than what they’re actually charging customers, I think all this math is going to work out in
0:21:17 the end.
0:21:23 And, of course, there’s an analogy to be drawn in kind of the early and mid-2010s where, you know,
0:21:30 companies like Uber, Airbnb, Lyft, these companies were just burning through VC capital, subsidizing
0:21:33 rides, subsidizing housing, and they made it, right?
0:21:36 And there’s a thousand examples of those that didn’t.
0:21:45 But the VC math is typically over-invest, capture the market, and then monetize more based on business
0:21:46 fundamentals later on.
0:21:47 Yeah, yeah, yeah.
0:21:51 I mean, Amazon, didn’t Amazon lose money for like 20 years or something like that?
0:21:51 Exactly.
0:21:55 And all they had to do was turn the dial, and all of a sudden, they had margins.
0:21:56 So it’s going to be interesting.
0:22:00 Maybe we’re in a mini bubble, but in the long run, I just don’t think we’re going to see this
0:22:03 huge drop like a dot-com bubble.
0:22:08 And there’s been a number of interesting tweets about this, where if you look at the price of
0:22:11 earnings, we’re nowhere near the dot-com era.
0:22:12 Yeah, yeah, yeah.
0:22:15 You were showing me that graph, I think, last Friday when we were doing a live stream together.
0:22:18 It’s really, really interesting to me.
0:22:22 Did you hear that, like, NVIDIA’s now got a stake in Intel, which is something that I never
0:22:23 saw coming either.
0:22:25 No, I didn’t see that.
0:22:30 I know the U.S. government has a stake in Intel, which is kind of nuts.
0:22:30 Yeah.
0:22:33 Well, that’s probably why NVIDIA went in, let’s be honest.
0:22:37 They probably saw the government got invested in Intel, and NVIDIA’s like, all right, we’re
0:22:37 in as well.
0:22:42 And, you know, I mentioned that GTC, the second version of GTC this year was in Washington,
0:22:43 D.C.
0:22:46 Well, I think they did that for, like, political reasons.
0:22:49 I think he did that to be close to Trump.
0:22:54 Obviously, Jensen, from what I understand during his keynote, was very, you know, favorable towards
0:22:56 Trump and how Trump is handling things.
0:22:57 Well, of course.
0:23:00 He seemed to be fairly okay with the whole China situation.
0:23:01 Yeah.
0:23:06 Look, I think Zuckerberg is probably the most prominent example of this, of just doing a
0:23:07 180.
0:23:11 These guys who lead the Mag 7, they are incredibly bright.
0:23:15 You know, ultimately, they’ll do whatever it takes to make the company successful.
0:23:18 And if that means kissing the ring of the current administration or whatever the next administration
0:23:20 is, they will do that.
0:23:23 And whether they believe it or not doesn’t really matter.
0:23:24 They’re going to do it.
0:23:29 And that is how they’re currying favor with the administration.
0:23:30 Yeah.
0:23:35 No, it’s been kind of funny, a little slightly sad to watch, but, you know, interesting nonetheless.
0:23:36 Expected.
0:23:37 Yeah, yeah.
0:23:42 I mean, seeing, like, Tim Cook hand that, like, clear disk thing to Trump.
0:23:47 And I think Jensen handed Trump, like, a signed, I don’t know, like, one of his signed GPUs
0:23:48 or something.
0:23:52 I don’t remember what he gave him, but he gave him something to get in his favor.
0:23:55 But yeah, no, it’s really interesting to see how this is all playing out, because I would
0:23:59 have thought that Jensen would really be fighting for the business in China, but he seems to be
0:24:01 okay just sort of letting that go.
0:24:02 Yeah.
0:24:05 Well, I assume he’s still going to be fighting.
0:24:09 I think China’s making it difficult to sell into China, though, because they see the writing
0:24:10 on the wall.
0:24:16 They know that if they build their AI infrastructure on top of U.S. chips and software, they are
0:24:20 going to be at a disadvantage and beholden to the U.S.
0:24:22 And of course, they don’t want that.
0:24:22 Yeah, yeah.
0:24:24 Totally makes sense.
0:24:29 Okay, let me tell you about another podcast I know you’re going to love.
0:24:34 It’s called Billion Dollar Moves, hosted by Sarah Chin Spellings.
0:24:38 It’s brought to you by the HubSpot Podcast Network, the audio destination for business
0:24:39 professionals.
0:24:45 Join venture capitalist and strategist Sarah Chin Spellings as she asks the hard questions and
0:24:49 learns through the triumphs, failures, and hard lessons of the creme de la creme, so you
0:24:53 too can make billion dollar moves in venture, in business, and in life.
0:24:59 She just did a great episode called The Purpose Driven Power Player Behind Silicon Valley’s
0:25:01 Quiet Money with Mike Anders.
0:25:04 Listen to Billion Dollar Moves wherever you get your podcasts.
0:25:11 Well, let’s talk quickly, too, about the OpenAI event that happened this week.
0:25:15 So they did sort of like an, I don’t know, what would you call the live stream?
0:25:17 A sort of state of OpenAI kind of update thing?
0:25:19 It felt like ad hoc.
0:25:19 I don’t know.
0:25:23 I mean, obviously, they had been planning it a while, but it just felt like out of nowhere.
0:25:27 And then, yeah, there was kind of a few minutes about the, like an overview of it.
0:25:28 It was kind of very technical.
0:25:31 Jakob Pachocki was there with Sam Altman.
0:25:34 Jakob being the chief scientist, I believe is his title.
0:25:35 And then they did a Q&A.
0:25:36 Yeah.
0:25:39 So Sam went to X and posted this TLDR.
0:25:42 I’m not going to read the whole thing because even his TLDR could use a TLDR.
0:25:47 Honestly, I think the first line, we have set internal goals.
0:25:51 If you want to read that, that is crazy important to me.
0:25:52 All right, let’s see.
0:25:58 So we have set internal goals of having an automated AI research intern by September of
0:26:04 2026, running on hundreds of thousands of GPUs and a true automated AI researcher by March
0:26:06 of 2028.
0:26:10 We may totally fail at this goal, but given the extraordinary potential impacts, we think it
0:26:13 is in the public interest to be transparent about this.
0:26:20 So in the talk of AGI, it sounds like Sam Altman has the belief that AGI is coming before
0:26:28 2028, but an automated AI researcher by March of 2028, like if AGI comes sooner than that,
0:26:30 wouldn’t the AI researcher also come sooner than that?
0:26:31 You know what I mean?
0:26:32 Okay.
0:26:36 There’s so much to unpack from this one paragraph here.
0:26:39 So first, he does not mention AGI.
0:26:41 And I actually think that’s important to your point.
0:26:46 They have ironed out a new relationship, a new dynamic, a new deal with Microsoft.
0:26:54 And a big part of the original deal with Microsoft was Microsoft had ownership of OpenAI’s IP up
0:26:56 until they hit AGI.
0:26:58 Now, AGI was such an amorphous term.
0:27:00 It was a moving goalpost.
0:27:04 Nobody truly knows what AGI means or how to measure it.
0:27:06 So kind of ignored that.
0:27:08 So there’s two other things I want to point out here.
0:27:14 One, these are incredibly specific dates for saying that they’re going to have something
0:27:18 like an AI researcher, an automated AI researcher.
0:27:19 September, 2026.
0:27:20 Okay.
0:27:24 That’s, you know, not too far away, but it is a very precise date.
0:27:30 The one that blew my mind is March, 2028 to have a true automated AI researcher.
0:27:34 It is a specific month within a specific year.
0:27:35 It just seems wild.
0:27:42 Like, what does Sam Altman know to have such a precise prediction of that timeline?
0:27:50 Now, the last thing that I want to point out is once a company has a true automated AI researcher,
0:27:57 he didn’t say any of this, but I’m going to infer that also means you’re going to have self-improving
0:27:58 artificial intelligence.
0:28:04 And once you have self-improving artificial intelligence, the only limiter for really just
0:28:11 an explosion of intelligence to use the situational awareness paper is how much compute can you throw
0:28:12 at it?
0:28:12 Right?
0:28:18 And so it seems to me, if you have these, you know, handful of companies racing towards
0:28:25 whatever that goal is, the self-improving AI seems to be really the only goal that matters
0:28:30 because once one of the companies hits that goal, they are forever in the lead.
0:28:35 And it just seems very obvious to me that I know Zuckerberg is heading towards that.
0:28:37 He calls it artificial super intelligence.
0:28:42 Sam Altman is calling it an automated AI researcher, but they’re all talking about the same thing.
0:28:46 They’re talking about this point at which AI can improve itself.
0:28:46 Right.
0:28:47 Right.
0:28:52 And to me, that’s also the point where, you know, AI has the potential to be the, you know,
0:28:54 quote unquote, runaway AI.
0:28:54 Right.
0:28:59 That’s the point where AI, if it’s self-improving, it might start to figure out like that’s where
0:29:04 the sort of paperclip maximizer theoretical experiments come into play is okay.
0:29:06 Well, I’m sort of optimizing for this goal.
0:29:08 So what do I have to do to achieve the goal?
0:29:11 You know, possibly at the expense of everything else.
0:29:13 That’s where things get scary.
0:29:16 The reason I bring up AGI is I know Sam Altman did an interview last year.
0:29:19 Maybe he was just being hype Altman.
0:29:19 Right.
0:29:23 But in an interview last year, when somebody said, what are you excited for in 2025?
0:29:25 He said, I’m excited about AGI in 2025.
0:29:26 Right.
0:29:27 Something along those lines.
0:29:29 He said, I’m excited about AGI next year.
0:29:33 So he almost implied last year that he thought we’d have AGI this year.
0:29:39 And now he’s saying, you know, the automated AI researcher by March of 2028, which seems like
0:29:42 if you have AGI before March of 2028.
0:29:46 Are those the same level, AGI and self-improving AI?
0:29:50 Like, would you see those as sort of like being the same thing or are those two different milestones?
0:29:55 So Matt, they seem, and based on my understanding, they are two completely different milestones.
0:29:58 Maybe not completely different, but they are very different.
0:30:03 And so when he says AGI, if we’re using his and open AI’s kind of previous definition of AGI,
0:30:12 it is essentially when AI can accomplish most economically valuable work better than the average or most humans.
0:30:15 I butchered it a little bit, but I think that was the definition.
0:30:16 And I actually do think we’re almost there.
0:30:19 Granted, we have two more months left in this year, so we’ll see.
0:30:25 But I think he got to a point with the Microsoft relationship where he couldn’t really use or
0:30:27 like have a hard definition of AGI.
0:30:34 Now, going towards self-improving artificial intelligence, a true AI researcher, that seems
0:30:35 very different.
0:30:44 That is when AI can discover novel algorithms and new ways to build silicon and then apply
0:30:46 it to itself, it becomes more powerful.
0:30:52 And thus, the next piece of research just happens more quickly or becomes, you know, a more important
0:30:53 discovery.
0:30:57 So I do think those are two very different things.
0:31:02 And AGI, in the context of his previous definition, does seem like it is right around the corner.
0:31:07 And in fact, a lot of these models, the core intelligence of the models, I think we’re there.
0:31:12 I just think we’re now needing to build the scaffolding and the implementation around it,
0:31:14 which seems to be lagging by about 18 months.
0:31:17 So the core intelligence of the model is there for AGI.
0:31:18 We just need to go implement it now.
0:31:22 So I don’t think he was wrong with his predictions.
0:31:25 It’s just the definitions that he’s using are pretty amorphous.
0:31:27 Yeah, yeah, I totally agree with that.
0:31:31 Yeah, I remember you were at Dreamforce this year doing your forward future live during Dreamforce.
0:31:35 And one of the terms I kept on hearing you bring up over and over again was capability
0:31:36 overhang.
0:31:37 Yeah, exactly.
0:31:40 It sounds to me like that’s kind of what you’re referring to here, right?
0:31:44 It’s actually smarter than what most people are using it for right now, right?
0:31:45 Yeah, exactly.
0:31:47 The models are extremely capable.
0:31:55 The scaffolding around them, the tools, you know, memory tool use, access to different information
0:31:58 sources, all of that stuff still needs to continue to be built out.
0:32:04 Yeah, along the same lines here, I just pulled up this article, both OpenAI and Microsoft posted
0:32:08 this article on their website about the next chapter of Microsoft OpenAI partnership, which
0:32:14 ties into what we’re saying, because they basically said once AGI is declared by OpenAI, the declaration
0:32:17 will now be verified by an independent expert panel.
0:32:23 So they’re going to put together some sort of panel to decide, OK, did OpenAI actually achieve
0:32:23 AGI.
0:32:28 So, I mean, it’s still subjective, but it’s now left up to the subjectivity of this panel
0:32:32 as opposed to like Sam Altman just claiming we got AGI.
0:32:33 Interesting.
0:32:35 So this just came out, right?
0:32:36 A couple of days ago.
0:32:36 Yeah.
0:32:37 Oh, interesting.
0:32:39 So maybe I got that wrong.
0:32:44 Actually, I thought the whole notion of you get the IP.
0:32:48 Microsoft OpenAI says you get the IP until we declare AGI.
0:32:52 But it seems like they still do have that kind of going up to this paragraph.
0:32:53 I can read real quick.
0:32:57 The agreement preserves key elements that have fueled this successful partnership, meaning
0:33:00 OpenAI remains Microsoft’s frontier model partner.
0:33:08 OK, Microsoft continues to have exclusive IP rights and Azure API exclusivity until AGI.
0:33:10 I might have made a mistake there.
0:33:16 I thought they had removed that part of the agreement, but it seems like they just clarified
0:33:22 that OpenAI alone doesn’t get to declare it, although they do get to declare when it happens.
0:33:25 They declare it and then an independent panel verifies it.
0:33:26 Right.
0:33:28 And so let’s look at the incentives here.
0:33:35 OpenAI is incentivized to declare AGI as soon as possible because they do not want to give
0:33:38 Microsoft exclusive IP rights, right?
0:33:40 They don’t want Azure to have exclusivity.
0:33:43 Any exclusivity is bad generally for a company.
0:33:49 And so OpenAI is definitely incentivized to claim AGI, and they could have done so at any
0:33:52 moment, but now they actually have to have a third party validate it.
0:33:52 Right.
0:33:54 And what that validation is based on, I don’t know.
0:33:55 Who knows?
0:33:55 Yeah.
0:34:01 But it does say here, Microsoft’s IP rights for both models and products are extended through
0:34:07 2032 and now include models post-AGI with appropriate safety guardrails.
0:34:13 So basically, it sounds like their agreement goes through 2032 regardless of AGI.
0:34:19 I’m immediately thinking of the Megamind meme and Satya Nadella just playing 4D chess.
0:34:21 This is what comes to mind.
0:34:24 I wish I could just instantly pull that meme up and show it right now, but you’re just going
0:34:25 to have to visualize it.
0:34:26 Yeah.
0:34:30 I mean, it seems like Microsoft worked out a pretty nice deal for this.
0:34:31 They had all the leverage.
0:34:32 I’m not surprised.
0:34:36 Microsoft’s IP rights now exclude OpenAI’s consumer hardware.
0:34:41 So whatever they end up making with Johnny Ive, if they ever do get into robotics, that’s
0:34:43 going to be outside of Microsoft’s scope.
0:34:50 I do remember there was a little bit of tension around when OpenAI approached WindSurf to acquire
0:34:50 WindSurf.
0:34:54 It seemed like it was Microsoft that got that deal shut down, right?
0:34:56 Like Microsoft went, oh, you’re buying WindSurf?
0:35:01 Well, because of our agreement, that means we’ll also own a portion of WindSurf.
0:35:07 And OpenAI didn’t like that, but it does sound like there’s no exclusions for like acquired
0:35:10 software from OpenAI, just consumer hardware.
0:35:11 Yeah.
0:35:14 And again, like Microsoft just had all the leverage here.
0:35:19 They invested and got all of these terms before OpenAI was OpenAI, right?
0:35:19 Yeah.
0:35:20 I think it was 2017.
0:35:25 And by the way, did you see Satya recently did an interview in which he said, Bill Gates
0:35:30 thought we were just burning a billion dollars on that investment in OpenAI, which is like,
0:35:32 you know, no, I didn’t see that.
0:35:32 Oh my God.
0:35:40 Even the most prominent tech minds in our multiple generations get stuff wrong.
0:35:42 Nobody really knows what’s going on.
0:35:43 Maybe except for Satya.
0:35:46 Satya seems to be just operating on another level.
0:35:47 He does.
0:35:47 Yeah, yeah, yeah.
0:35:53 Also, Microsoft can now independently pursue AGI alone or in partnership with third parties.
0:35:58 So I mean, OpenAI and Microsoft are competitors now in the race to AGI.
0:35:59 Oh, absolutely.
0:36:02 They’re going to be competitors on almost every dimension, right?
0:36:03 OpenAI is building hardware.
0:36:09 It’s just like a completely new way to write and build software.
0:36:15 They can compete with every piece of Microsoft software all the way down to the operating system
0:36:15 level.
0:36:16 I truly believe this.
0:36:21 Microsoft sees OpenAI as an existential threat, but they’re also a partner.
0:36:24 They’re like, you know, keep your enemies close type thing.
0:36:25 Yeah, no, absolutely.
0:36:28 So let’s talk about the robots real quick here.
0:36:29 Let’s talk about robots.
0:36:34 We got the Neo home robot that was announced this week, which you can actually pre-order
0:36:35 for a $200 deposit.
0:36:37 Did you see this one at CES?
0:36:41 They actually had this one at CES and they were doing like a demo of it, like vacuuming
0:36:42 a living room kind of thing.
0:36:43 No, I think I missed that.
0:36:49 This one is just interesting because it’s like a robot wearing clothes to kind of hide
0:36:50 the fact that it’s a robot.
0:36:54 So do you want to quickly fill me in on what’s got you excited about the Neo robot?
0:36:55 Yeah, absolutely.
0:37:03 So this is really like the first time a company has officially opened orders for a robot that
0:37:08 you can buy and have in your home today, with the exception of Unitree, which is a Chinese
0:37:09 company.
0:37:12 I was going to say, I’ve seen like streamers and stuff that have like the Unitree robot.
0:37:15 Like, I don’t know if you saw the Kaisenat one where they were like pushing it around and
0:37:15 stuff.
0:37:16 It was awesome.
0:37:16 Yeah.
0:37:20 So yeah, it’s definitely possible to buy it, but you’re going to be buying it from China
0:37:22 and everything that comes with that.
0:37:28 So this is the first US company that you’re able to actually buy a humanoid robot that is
0:37:29 supposed to do chores around your house.
0:37:31 It is built for your house.
0:37:36 They show off a bunch of examples, cleaning, welcoming guests by opening the door, doing
0:37:41 laundry, cooking food, potentially getting medication for you.
0:37:45 I mean, there’s a thousand use cases and their launch of video, as I mentioned earlier, went
0:37:46 absolutely mega viral.
0:37:52 50 million plus views just on X alone, not including YouTube and Instagram and everywhere else they
0:37:53 probably launched.
0:38:00 Now, Joanna Stern, I believe her name is, actually had an in-person review and demo of this robot.
0:38:04 Turns out it is 100% tele-operated.
0:38:10 And what that means is there is somebody wearing VR goggles with handset of some sort operating
0:38:10 the robot.
0:38:15 So it is not yet true artificial intelligence inside the robot.
0:38:19 Now, that is not to take anything away from this accomplishment.
0:38:24 It is wild to think just a few years ago, this was a complete pipe dream.
0:38:29 And now we actually have working humanoid robots that seem to be quite impressive.
0:38:37 MKBHD just launched a video yesterday talking about how a lot of tech companies are now over-promising
0:38:38 and under-delivering.
0:38:40 This is kind of another prime example.
0:38:47 And then even like within tech, within AI, it’s even more prominent that companies are over-promising,
0:38:47 under-delivering.
0:38:50 Of course, he was talking about the rabbit and-
0:38:51 Humane.
0:38:52 A humane pin, right.
0:38:57 And so once again, the demos so far today showed that it is 100% tele-operated.
0:39:02 In the demo video, there were two different scenes in which it was fully autonomous.
0:39:03 That’s great.
0:39:07 But the promise is it is the majority of uses is going to be autonomous.
0:39:11 They said they’re going to be delivering this by the beginning of next year.
0:39:11 Yeah.
0:39:16 So it says right here, US deliveries begin in 2026, starting with early access customers.
0:39:18 It doesn’t say when in 2026.
0:39:21 It just says be in 2026.
0:39:24 So they basically have over a year to start delivering this.
0:39:26 I put in a pre-order, by the way.
0:39:26 I’m super excited.
0:39:28 I tend to be very optimistic.
0:39:31 I like thinking that this stuff is going to work.
0:39:33 But of course, they got the robot.
0:39:34 It works great.
0:39:35 It moves well.
0:39:38 But that last 10% to 20% always is the hard part.
0:39:41 And I think it’s very analogous to self-driving.
0:39:46 Elon and Tesla have been talking about delivering fully capable self-driving for over a decade.
0:39:53 And we’re now just kind of like starting to get to somewhere that it feels like, okay, he’s delivering on that promise.
0:39:57 But it was, you know, a year away at all for the last decade.
0:39:58 It seems really cool.
0:39:59 I’m into it.
0:40:08 The part that is maybe a little worrisome for me is if I have this humanoid robot in my house and it’s not able to do things autonomously.
0:40:13 There’s going to be somebody else operating this robot, seeing into my house, doing things in my house.
0:40:17 And that does not rub me the right way.
0:40:18 Interesting.
0:40:18 Okay.
0:40:20 I didn’t catch that when you said it the first time.
0:40:23 So I understood that they’re tele-operating it, but I assume that was for the demos.
0:40:29 But like if you get one of these and put it in your house, somebody is going to be tele-operating it while it cleans your house?
0:40:30 Yeah.
0:40:31 So it’s in the FAQs.
0:40:48 But basically what happens is if you give it a task and it is not able to accomplish it autonomously, which most likely is going to be a lot of tasks in the early days, you can schedule a time with their tele-operators for them to accomplish the task.
0:40:57 Now, the hope is by the tele-operator accomplishing the task through Neo, Neo will learn and not need a tele-operator for subsequent tasks.
0:41:08 But again, you are giving permission for this person running this robot that’s capable of lifting 150 pounds, by the way, to operate in your house, which is just, it’s wild.
0:41:14 I don’t know if I feel more uncomfortable with having an autonomous robot doing that or somebody operating it.
0:41:15 I think it’s the latter though.
0:41:16 Yeah.
0:41:24 I think somebody operating it is creepier to me that like somebody is looking around in my house, like know where I put my wallet, things like that.
0:41:24 Yeah.
0:41:25 Yeah.
0:41:32 But, you know, on the flip side of the coin, if you’re looking at like one silver lining, you know, everybody keeps on saying AI is going to create new jobs.
0:41:37 This is a new job that didn’t exist, tele-operating robots for people.
0:41:38 Yeah.
0:41:38 Yeah.
0:41:40 Matt, what do you think about the look?
0:41:43 What do you think about the design of the robot?
0:41:49 I think, honestly, it’s one of the less cool robots that I’ve seen.
0:41:51 What makes you say that?
0:41:51 I don’t know.
0:41:55 To me, it looks like a giant human-sized teddy bear or something.
0:42:06 Do you remember like a few years ago, Elon Musk went on stage to announce Optimus and they had a bunch of people with like Optimus outfits on dancing?
0:42:06 Yeah.
0:42:09 It feels like that, but there is actually a robot under it.
0:42:18 For me, like having the very soft look wrapped in cloth and just, it looks a little, I don’t know.
0:42:21 It’s not my design aesthetic that I would choose.
0:42:25 When I see figures robot, that is what I want robots to look like.
0:42:29 I want to see the metal and the different materials.
0:42:29 I agree.
0:42:30 That’s where I’m at too.
0:42:34 This to me looks like it’s from like a Tim Burton movie or something, if I’m being honest.
0:42:36 Oh, yeah, totally.
0:42:38 But anybody who’s wondering, I just looked up the price here.
0:42:44 So you can buy one for 20 grand or there’s a subscription for 500 bucks a month.
0:42:50 So like if you can’t afford the 20 grand up front, you can basically like rent one of these robots to live in your house.
0:42:54 So, Matt, let’s assume this is all going to work out.
0:42:56 It’s going to deliver on its promise.
0:42:58 Whether that happens or not, let’s make that assumption.
0:43:01 The 20K price point is very interesting.
0:43:08 That is essentially the lowest priced car in the United States is a little bit, I think it’s 18,000.
0:43:12 So it is essentially a very inexpensive car.
0:43:21 Plus, if you start layering on financing options, it really becomes something that I think a broad portion of the population can afford.
0:43:23 500 bucks a month.
0:43:35 OK, that seems like a lot of money, but imagine how much time you will save not having to clean up, not having to do the dishes, not having to tidy.
0:43:37 And I guess all of these use cases are really cleaning.
0:43:42 But I just think that the price point was the most compelling aspect of this launch.
0:43:52 Yeah, well, I mean, I figure if you hire like a housekeeper, right, like let’s say they come once a week, you’re probably going to pay 500 bucks a month for that housekeeper to come once a week.
0:43:56 This is someone that something that is there every day.
0:43:56 Right.
0:44:01 So if you were to hire a private chef, right, that’s a whole additional cost.
0:44:03 If you were to hire a housekeeper, that’s an additional cost.
0:44:13 Like if you were to hire out all of the individual things that this one robot theoretically will be able to do for you, this is actually the more economical option, it seems.
0:44:14 Did you preorder?
0:44:15 I haven’t yet, but I probably will.
0:44:17 I mean, it’s 200 bucks.
0:44:20 I tend to preorder pretty much every new tech gadget that comes out.
0:44:21 So I had a feeling.
0:44:22 Yeah, yeah.
0:44:23 I put in the preorder.
0:44:29 I’m going to go with the early access because that’s what I want and I want to show it off on camera.
0:44:40 So, uh, yeah, my Neo robot, after I learned the third world or piloting them is mad chill in video games with them in the house is a mess.
0:44:41 Excellent.
0:44:42 Just hanging out.
0:44:42 Yeah.
0:44:44 What do we got here?
0:44:46 Uh, day two of cleaning after humans.
0:44:47 Yeah.
0:44:49 My robot is going to see things.
0:44:50 Let’s just say that.
0:44:52 Yeah.
0:44:52 We’ve got kids.
0:44:54 There’ll be, uh, yeah.
0:44:58 Our robots are most likely to turn on humans because of what they have to see.
0:44:58 Yeah.
0:45:01 Uh, but yeah, this is hilarious.
0:45:05 I think this is a whole new category of meme and I’m ready for it.
0:45:06 Yeah, absolutely.
0:45:09 Well, Matt, this has been absolutely amazing.
0:45:13 I love having you on because you and I can nerd out about AI forever.
0:45:16 I feel like every time we have a conversation, it always ends the same way.
0:45:18 I feel like we could talk about this stuff forever.
0:45:18 Oh yeah.
0:45:20 We could easily keep going for another two hours.
0:45:21 Next time.
0:45:26 But, um, well, we’ll definitely have you back on if you’re willing to come back on and do this again.
0:45:27 Absolutely.
0:45:31 But it’s been a blast and, uh, I’m really loving what you’re doing with the forward future live.
0:45:31 Thank you.
0:45:35 I love those live streams that you’re doing and you’re rotating through guests and stuff.
0:45:35 Really, really, really.
0:45:35 Yeah.
0:45:37 We have to have you back on speaking.
0:45:37 Yeah.
0:45:38 It’s a blast.
0:45:39 It’s a blast.
0:45:39 So thank you.
0:45:40 Thanks so much.
0:45:41 Any place you want to shout out.
0:45:44 Obviously, everybody needs to check out Matthew Berman on YouTube.
0:45:45 Check out forward future.
0:45:48 That’s where he does the live show forward future live as well.
0:45:50 Anywhere else people should go after listening to this.
0:45:51 Yeah.
0:45:55 So if you can just click on the forward future Twitter thing right below Nick Wentz on the
0:45:59 right side of the screen, just follow us at forward underscore future underscore.
0:46:04 We’re working on getting a better handle, but for now it is that follow us at forward future.
0:46:04 Awesome.
0:46:05 Well, Matt, thanks so much.
0:46:07 And, uh, looking forward to doing it again with you in the future.
0:46:08 Thanks, Matt.
0:46:48 Thank you.
Want Matt’s favorite AI tools + playbook? Get it here: https://clickhubspot.com/vgb
Episode 83: Are Adobe’s new AI tools the future of creative work, or could generative models spell the end for legacy platforms like Photoshop? Matt Wolfe (https://x.com/mreflow) is joined by Matthew Berman (https://x.com/MatthewBerman), creator of Forward Future and a leading voice covering the front lines of artificial intelligence, from major tech events like Dreamforce to hands-on interviews with the innovators shaping tomorrow.
In this episode, Matt and Matthew break down the biggest headlines from the week in AI: Adobe’s conversational assistant and existential business challenges, Nvidia’s mind-bending new investments and political maneuvering, OpenAI’s bold timeline to build a self-improving AI researcher, and the viral Neo Humanoid robot—are we ready to trust a home robot with our privacy? Packed with fresh takes, inside scoops, and speculative predictions, this fast-moving conversation is your front row seat to the unfolding era of AI and robotics.
Check out The Next Wave YouTube Channel if you want to see Matt and Nathan on screen: https://lnk.to/thenextwavepd
—
Show Notes:
-
(00:00) AI Insights and Future Predictions
-
(03:41) Photoshop Adds AI Chat Assistant
-
(08:08) Adobe, AI, and Creative Future
-
(10:26) Adobe’s AI Future Concerns
-
(15:53) Nvidia GTC Highlights
-
(17:23) Nvidia’s Investment Cycle Explained
-
(20:23) AI Investment: Over-Investing Now
-
(26:06) Automated AI Researcher Timeline
-
(29:00) AGI vs Self-Improving AI
-
(30:47) AGI Verification Panel Announced
-
(36:04) First US Humanoid Robot Launch
-
(39:18) Robot Tasks: Autonomy vs. Operators
-
(41:50) Affordable Car with Practical Benefits
-
(44:07) Future Live Streams Enthusiasm
—
Mentions:
-
Matthew Berman: https://www.linkedin.com/in/matthewberman
-
Forward Future: https://www.forwardfuture.ai/
-
TechCrunch Disrupt: https://techcrunch.com/events/tc-disrupt-2025/
-
Nano Banana: https://nanobanana.ai/
-
Nvidia GTC: https://www.nvidia.com/gtc/
-
Neo Humanoid Robot: https://www.1x.tech/order
Get the guide to build your own Custom GPT: https://clickhubspot.com/tnw
—
Check Out Matt’s Stuff:
• Future Tools – https://futuretools.beehiiv.com/
• Blog – https://www.mattwolfe.com/
• YouTube- https://www.youtube.com/@mreflow
—
Check Out Nathan’s Stuff:
-
Newsletter: https://news.lore.com/
-
Blog – https://lore.com/
The Next Wave is a HubSpot Original Podcast // Brought to you by Hubspot Media // Production by Darren Clarke // Editing by Ezra Bakker Trupiano
Leave a Reply
You must be logged in to post a comment.