Anthropic’s Potential, AI Video Gets Crazy, Market Bubble Approaching?

Ranjan Roy from Margins is back for our weekly discussion of the latest tech news. We cover: 1) Alex's profile of Anthropic CEO Dario Amodei + extra reporting 2) Are Anthropic's convictions earnest or marketing? 3) Where does Anthropic fit in the broader AI landscape 4) Will one AI company 'win it all' 5) Anthropic is raising at a new, $170 billion valuation 6) OpenAI hits 700 million ChatGPT users 7) Hyped for GPT-5? 8) AI video creation's promise 9) Videos in Google's NotebookLM 10) Is the stock market in a bubble or hitting real returns. --- Enjoying Big Technology Podcast? Please rate us five stars ⭐⭐⭐⭐⭐ in your podcast app of choice. Want a discount for Big Technology on Substack + Discord? Here’s 25% off for the first year: https://www.bigtechnology.com/subscribe?coupon=0843016b Questions? Feedback? Write to: [email protected] Learn more about your ad choices. Visit megaphone.fm/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Alex Kantrowitz?

  This episode is brought to you by Genesis. What does it actually take to put Agentic AI to work across an entire enterprise? I recently spoke with Genesis chairman and CEO Tony Bates about why orchestration is emerging as a competitive advantage in customer experience and what companies need to get right in the AI era. Then I sat down with Adam Mitchell, head of enterprise business solutions at Voya Financial, to go inside a large-scale Agentic CX transformation, including the business decisions, trade-offs, and lessons along the way. You can watch both conversations now on my YouTube channel today. It's never been more challenging to get frontier compute. How do you know where to look, who to trust, or how much to pay?

• 0:43Alex Kantrowitz?

  Akash is an open cloud that provides easy access, flexibility, and peace of mind. That's how companies like Razer were able to cut their inference cost by 15 times. Visit akash.network/alex for a free consultation. That's akash.network/alex for a free consultation.

• 1:03Alex Kantrowitz?

  Anthropic is raising at a new $170 billion valuation. Let's talk about the company and its potential. Plus, OpenAI hits 700 million ChatGPT users. Is AI video actually underhyped? And is the stock market in bubble territory? That's coming up on a Big Technology Podcast Friday edition right after this. Welcome to Big Technology Podcast Friday Edition, where we break down the news in our traditional, cool-headed, and nuanced format. We have a great show for you today. We're going to talk about Anthropic's new massive funding round and big valuation. We'll also cover my interview with Dario Amode this past week and talk a little bit about the takeaways there, where Anthropic is positioned, and what the company's promise is.

• 1:44Alex Kantrowitz?

  We'll also talk about OpenAI's 700 million ChatGPT users, a great section on AI video, and whether that format is underhyped, and also the stock market rip-roaring. Microsoft joins Nvidia as a $4 trillion company. Is it in bubble territory? Joining us as always on Fridays to do it is Ranjan Roy of Margins. Ranjan, great to see you. Welcome back to the show.

• 2:05Speaker 01

  Good to see you. I missed you last week, but it was nice to listen to the Dario episode on my plane ride back from Tokyo. So excited to talk about this.

• 2:13Alex Kantrowitz?

  Okay, great. So on that note, I just want to welcome all of our new subscribers who've come in. We have a bunch of new people here. Just to give you a rundown of what we do here on Wednesday, we have a big flagship interview with people like Dario. Next week, I'm Jav Masad, the CEO of Replit is coming on. And then on Fridays, Ranjan and I show up here and break down all the week's tech news. So we're very, very glad to have you here. All right, Ranjan, I actually thought it would be good to start this week just to talk a little bit about my takeaways after doing a couple of months of reporting on Anthropic, and then we can get into their fundraising. So let me just start here. I think that Anthropic is a very impressive company. Like, the one thing that I kept coming back to when writing the story about Dario and the company is just that you have this competition coming

• 2:57Alex Kantrowitz?

  from all sides, competition, obviously, from OpenAI, but also from Meta, places like Google, or DeepMind. And Anthropic has found a way to sort of make its way into this conversation, starting really later than OpenAI and without the resources of the big companies. And I think there was a great quote from Eric Schmidt that I didn't include in my story. I had a conversation with the former Google CEO for this profile of Dario. And he says this, he says, "My best description of this is that it is an all-out brawl. It's really a brawl between very well-capitalized companies, including my friend Dario's startup. The fact that Dario is in the brawl is extraordinary. And if you ask me what I predicted, if I predicted he'd get there, the answer is no. How would he raise the money? How would he

• 3:42Alex Kantrowitz?

  get the partnerships? And somehow he did." And I think it is interesting, despite the fact that Anthropic has 11 billion plus dollars in funding from big tech, that it's been able to compete in this market. What do you think?

• 3:54Speaker 01

  No, I think I love this all out brawl. I guess I'm not as surprised that Anthropic has been able to kind of compete and raise in this market because again, from a product perspective, we were Claude Boys back in the day, Claude Pro, maybe not subscribing right now, but they made that decision.

• 4:13Alex Kantrowitz?

  We were Claude Heads and Big Boys.

• 4:15Speaker 01

  Claude Heads, Claude Heads, Big Boys.

• 4:16Alex Kantrowitz?

  These are important details.

• 4:18Speaker 01

  These are, I thrive on accuracy here, but I think the most impressive thing that Anthropic has done is actually pivot into where they see opportunity. We've been talking about this the last few weeks and I'm curious kind of your takeaway after the conversation around Anthropic is making the pivot towards coding, software engineering, going after technical folks. And I think they're good, they're strong. Everything you see about Claude Code, and this is not me, which is why I'm not using it as much anymore and I can tell, but finding that very large market opportunity and niche and trying to win there rather than trying to do everything, I think they're pretty impressive right now.

• 5:02Alex Kantrowitz?

  So I think when you're not as well capitalized as a company like Meta, for instance, you have to make really strategic decisions. And I think speaking with Dario and those around him, I think Anthropic has made a couple of really good decisions in terms of where it's going to compete, really strategic areas in terms of where it wants to build its models and its businesses. So first of all, I think one interesting thing Dario said was that they are all in on the business use case for two reasons. I think maybe the first one, I'll just impute a little bit, is that if you're very good at business, you're going to make a lot of money and you need money to run in this type of industry. But the thing that Dario said that was interesting is that if you focus on business use cases, you have an incentive to make your models a lot better. For instance, if you're building

• 5:43Alex Kantrowitz?

  a chatbot, you just need to sort of cater to the standard uses. And what he said was basically that if you make your model move from, let's say undergraduate biology to graduate student biology level, the average chatbot user isn't going to see that, but the average business customer in, let's say, the pharmaceutical industry really will. And so that incentivizes them to build better models and it's also good business. And I just found that to be really smart.

• 6:10Speaker 01

  Well, so it's clear Dario's team model here for new listeners.

• 6:14Alex Kantrowitz?

  He is the most hardcore of all on team model, but go ahead. Yeah.

• 6:18Speaker 01

  For newer listeners, a regular debate between Alex and myself is I still believe the model is important, but the overall product and experience actually is where you see the real differentiation. So that's interesting to me that that is how they're kind of approaching this because remember Claude artifacts predated ChatGPT canvas like they kind of invented that. So to me, they've, they have innovated on the product side. So it's interesting to me that he's still all in on the model is where this brawl is going to take place.

• 6:51Alex Kantrowitz?

  Definitely. And it's also where you direct the model. So I asked him about the coding use case, which, as you mentioned, is an exploding use case for Anthropic, although we'll see what happens with GPT five and we're going to get into that in a bit. But his answer on coding was also really interesting. And that was that A, if we build great coding tools, every, all the engineers in Anthropic are going to use them, we're going to get faster as a company. And B, if you provide a useful AI tool to engineers, they're going to adopt it really fast. And that's proven to be the case. So it's not just the model, it's sort of the way that you train the model and how that outputs in product. And I think that's also a very wise bet for Anthropic that's paying off.

• 7:29Speaker 01

  Yeah, I mean, I think that's the success of Cursor, like the whole, you know, epithet that something is a wrapper has always been wrong to me. The Cursor skyrocketed because it just was a better experience. It provided better output relative to the competition at the time, but overall for software developers, it just was an overall better way to develop and code and engineer. So, so I think it, I think that, yeah, the two are inextricably linked, but, but I'm curious, what did you take away in terms of where Anthropic is going? And we're going to talk about the funding in just a moment. Clearly there's momentum, but, but what does he see, really see as the future in your takeaway?

• 8:12Alex Kantrowitz?

  I mean, I think that one thing that's clear is that Dario, this sort of does link to the funding, believes that he's the firmest believer in the scaling laws, which is that if you add more compute data and increase the model size, you're going to get predictable exponential results. And so I think that's basically the way that they want to move is they have some near term business interests to get them to a point. Where they can build bigger models and more efficient models and effectively trying to build a supermodel like I think everybody is like it's one thing that's going to come up I have MG Seidler coming on we have MG on the first Monday of the month and he writes this great newsletter called Spyglass and we're going to talk about the fact that all these companies are effectively working towards the same thing like call it superintelligence

• 8:56Alex Kantrowitz?

  call it you know AGI effectively they're working on scaling up large language models and adding sort of the scaffolding around them to make them useful so I think that's really where they want to go is they believe that they can build the models best they want to raise the money and build the biggest and best models and then

• 9:14Speaker 01

  See what happens.

• 9:15Alex Kantrowitz?

  What happens from there but we've seen these predictions from Dario that he has a belief that there's going to be like real economic value there and also some like potentially scary outcomes he became somewhat famous for saying that AI in the next couple of years could wipe out 50% of white collar entry level jobs I don't agree with that maybe that's a little bit of marketing for anthropic like we talked about but I really think labor augmentation and labor replacement is going to be where their business is.

• 9:42Speaker 01

  Well after the interview do you feel that was marketing or did you walk away that he truly believes this is worried about it is trying to do something about it?

• 9:52Alex Kantrowitz?

  So yeah and I'll note it wasn't just the interview there were 24 plus interviews that I did leading up to this story about him I do believe that yeah he both does really truly believe this he's seen these scaling laws work in places like Baidu and OpenAI and now at anthropic the data that Dario is looking at all points to this continuing to improve and in fact like Demis Hassabis from DeepMind said yes scaling is really helpful but we're seeing some diminishing returns or the word that he uses it's slowing and the people that I spoke with at anthropic like Jared Kaplan the chief scientist there said no it's not if scaling is slowing it's just because you're building it wrong most

• 10:36Alex Kantrowitz?

  probably so I really think that they do really believe in this stuff that this is going to continue on this exponential I believe that's an earnestly held belief I also think that it's undeniable that being loud about that is great marketing because if you're like the firmest believer in where the technology is going to go and you're talking about like You know how it's, you know, potentially like capable of bio, uh, bioweapons than if you're a bio, you know, if you're a pharmaceutical company, you're going to be like, well, I'm talking with some guy, I can't leave these anthropic guys out of the conversation.

• 11:10Speaker 01

  They're making bioweapons. I'm just trying to make, uh, the next Ozempic here.

• 11:14Alex Kantrowitz?

  Exactly. Ozempic plus. Lose, lose twice the weight in half the time. Yeah, I'm going to use AI for that.

• 11:20Speaker 01

  I'm in, I'm in.

• 11:21Alex Kantrowitz?

  But yeah, so, so that is, so it can be both. I really think it's both. I think everybody around who's been around them, you know, pro or con thinks that it's both. But I think the big issue, and this is sort of like takeaway number two for me, is that there's going to be an issue with inference and running these models is very expensive. Many of the startups who I've spoken with, uh, have said that like they like using anthropics models, but they're down too often. As the models get bigger, right? Cause it like, if the scaling laws are a thing, the models are going to get bigger. They're going to be more expensive to run. And if they're more expensive to run, are they more expensive to use? And if they're more expensive to use, does this industry work? I think that's the big open question. Here's Dario. He says, "We make improvements all the time, uh, that

• 12:04Alex Kantrowitz?

  make the models 50% more efficient than they were before. We're just at the beginning of optimizing inference. Inference has improved a huge amount from where it was a couple of years ago to where it is now." And I think whether they're able to continue that trend of improving inference and making it more efficient, that's the key question, or another key question, shall we say here? Because that, if you're building for the business use case, you have to make your model fit into their ROI calculations. And that's really the bottom line.

• 12:32Speaker 01

  Yeah, I think that is, the more I've been thinking about it, it, it has to become more of a conversation because right now, again, you know, most of these companies are just bleeding money right now. The economics of it, as you try to make the model more powerful and powerful and saying that's how you're going to get to the eventual productivity that is promised. But in the meantime, no one has modeled out what this costs a business or even an individual. We're all, I mean, this is like the mid 2010s when we're all getting free Uber rides. Now we're all just getting, as I was traveling around Asia, I'm sure what chat I was costing ChatGPT in all my searches.

• 13:12Speaker 01

  Was not what I was paying for for the 1995 or whatever it was for ChatGPT, 20 bucks for ChatGPT Plus. So I think the economics of the industry, and you had Ed Zitron on a couple of weeks ago. I firmly stand differently than him in terms of, I do believe generative AI and AI is going to completely transform everything. But what that looks like, what it costs, what cost structures look like for people and businesses. I, I think like this, it's certainly not been figured out yet.

• 13:45Alex Kantrowitz?

  Yeah. And by the way, like in the conversation, again, just to point back to it. I brought that up to Dario and was like, what's going on? You're going to potentially lose money on your most enthusiastic customers. Uh, and he said, well, that's going to change. And then of course, uh, this week Anthropic introduced stricter rate limits for Claude code users. So the company is real, and I think all AI companies at this point. They're trying to, or they're going to have to navigate the boundaries between like, you know. Yeah, it's like the opposite of social media, social media. You want people to be like super engaged and never leave your product. And with generative AI, because it is expensive to run, you have to be careful about the amount of use you try to sort of incentivize.

• 14:27Speaker 01

  Yeah, I mean, we've talked about this plenty. Generative AI is not traditional software economics. It's not like essentially zero marginal cost. It's not, certainly not social media, but not even traditional software. It's a completely different cost structure. It's more industrial, in my opinion, than traditional software. So it, which is actually, I would say like intellectually, one of the more interesting things for me that what the entire world is going to look like in terms of how it leverages generative AI, what these companies are going to look like. And to me, even for Google, that is actually one of the more interesting parts that the entire business has been built around.

• 15:07Speaker 01

  Kind of like fast adoption, slow, like Gmail and Google works space in the past was just get it in for anyone's hands and start building out more experiences and give them as much usage as possible. Whereas now Gemini is, is different and, and you can see that in the way they're all trying to figure out how to price this stuff.

• 15:27Alex Kantrowitz?

  Can I say we have a pretty intense debate going on in our big technology discord about this? There are like two sides of this equation. One side is that just get the usage and all costs will fall to zero because there's so much investment in this space. And the other is talking about how like, you know, I'm losing money on every product I make. I'm just making that up in volume. So, you know, we will see.

• 15:50Speaker 01

  Tale as old as time.

• 15:52Alex Kantrowitz?

  Tale as old as time, yeah. But I think that's one of the big open questions that I walked away with.

• 15:57Speaker 01

  Yeah, no, agreed. When venture is involved, that's the eternal debate.

• 16:01Alex Kantrowitz?

  Exactly. All right. Let me give one last quick thing. And that is that one of the main things that I wanted to go into this interview and this conversation was trying to investigate whether anthropics rise in revenue. We all know they've gone from like one billion last year. They're projecting nine billion this year, which we're going to get to in a moment. I just wanted to know, is the thing that is buoying the revenue, is that all coding? Or is it broader? And I learned it's much broader. So they are working with companies and industries like travel, healthcare, financial services, insurance, companies like Pfizer, United Airlines, and AIG to implement this technology within their workflows. And it's not just coding. So I think that the fact that it is broader than coding is a bullish sign for the company.

• 16:48Speaker 01

  Yeah, I mean, regular listeners will know, this to me is the ultimate battleground is is actual business implementation and use cases and within the enterprise. And for all the talk, this is not real yet, especially for the anthropics of the world. So I think they're clearly moving into the space. But to me, I don't know, like, they have these use cases, they're starting to build out their teams in these spaces. But on the other hand, like, does that decrease focus from the pure software development where they could win? And I think that's the question all these companies are going to have to face. They have to do everything. Again, we're going to talk about the funding in

• 17:29Speaker 01

  just a second, but they will anyone truly is this winner takes all? Or will certain competitors, certain people win niches within this space?

• 17:40Alex Kantrowitz?

  That is a great question. So I'm curious what your answer is to that and where you would put Anthropic in sort of the ranking of these companies.

• 17:48Speaker 01

  I'm a firm believer that no one is going to win it all. The consumer and enterprise are different, that a different, you know, like the act of software engineering is different than the act of marketing. And, and I'm definitely not a god model. One model will win and be able to do everything, do everything well. And that company will also have the best product for all types of use cases and for everyone. So, so I definitely think like, that is not going to happen. I mean, we've seen in tech, like, social media for a long time, Facebook dominated, then you start to see some level of competition. But, but the network effects within something like

• 18:30Speaker 01

  social media, I think are very different than the network effects from like scaling.

• 18:36Alex Kantrowitz?

  Let me ask you this. Is the risk that there are some economically clearly economically valuable areas of this world, and all these big model companies end up directing their attention there, and then splitting the market, and then just not having the return on investment that they expected? Let me just give coding as an example. Anthropic obviously has the lead in coding. It's been a good business for them. Google just brought the leadership of windsurf in, apparently in an attempt to get better at AI coding. GPT-5, which we're going to talk about in a moment, is expected to be much better at coding. If you have these three multi-billion dollar efforts going at the same market, dividing it up, is it a, is it potential, is there a

• 19:21Alex Kantrowitz?

  potential risk that like, no one wins, right? Because there's not going to be one taking all. Is it possible that you just end up getting this like spoiler effect where like, nobody really sees the return they want?

• 19:34Speaker 01

  I think you just defined free market capitalism and competition, Alex, and I like that. I think that's a good thing. I think by no one wins, this just goes back to the question around the econo- what do the economics of this look like eventually? And to me, Mark Zuckerberg doling out 100 million per engineer, which we talked about last week and said it is justifiable. But I think we start to move away from that arms race mentality and people have to hire sales teams and go to market teams and build good product and keep like prevent churn. And we just get back to the normal way business works. And you started

• 20:16Speaker 01

  this episode saying, if you're good at business, you make money. And I think if there's ever a takeaway, if you're good at business, you make money. And I think that's what it'll come back to, rather than the idea, given the rate of competition, that it will make the idea that any one company is going to be so far ahead purely from a model standpoint, that you'll have to win on the other parts of, again, like Salesforce didn't win cloud early on because it just was the best at cloud. It won cloud because it was just like a powerhouse and the rest of doing business. So I think, again, this goes back to like

• 21:01Speaker 01

  social, like Facebook, even Google, they created software monopolies solely on incredible software. But these types of businesses have to have the rest of the, like, Google wouldn't have, without its advertising model, wouldn't have been that interesting. So it's still not, it's still going to require more.

• 21:22Alex Kantrowitz?

  That's right. And just as that requirement shows up, we're starting to see the stakes increase again. So this is from Bloomberg Anthropic New Year's Funding at $170 billion value as revenue searches. The investment is going to be, they want to raise as much as $5 billion in a new round. By the way, they just raised most recently in January. They've been in discussions with iconic capital to lead the Qatar Investment Authority and Singapore Sovereign Fund about participating. And then the company was generating about $4 billion in annual recurring revenue or annualized recurring revenue earlier this month. But it now expects that revenue could hit $9 billion by the end of the year. This is a new stat from the Bloomberg story.

• 22:07Alex Kantrowitz?

  One thing that stuck out, Dario was like, listen, I am surprised, but we've seen revenue 10x from 0 to 100 million, from 100 million to a billion. And from 1 billion, we're now at 4, he said 4 billion annualized or maybe 4.5. So they're going to have to like, seriously, like the run rates are going to have to look like they're in the 20s, I think, at the end of this year, if they're going to hit 9 billion. Because they're going to have to make up for those months that they weren't on the $4 billion run rate. I don't know, maybe it's possible. They're going to get the money to find out.

• 22:42Speaker 01

  What are they going to do with the money? That's again, going back to, is it, I'm so curious and I know even with 24 interviews and diligent reporting, getting a true answer of how does that investment look like on a proportional level between investment on the model side and investment on the like business side, let's say, is a, is an answer that's almost impossible to get. But I don't know. Did you get any sense around that?

• 23:09Alex Kantrowitz?

  Oh yeah. I don't think it's that hard to figure out. I think it's going to go towards infrastructure and getting larger data centers.

• 23:17Speaker 01

  So yeah, all right. You just answered the question. Yep.

• 23:20Alex Kantrowitz?

  I mean, I think that's one of the things that's coming up here and why Mark Zuckerberg has been offering all this money. It's actually like a pretty smart thing. It's like, all right, well, if I'm spending, you know, 100 billion on data centers, I'm just like throwing a number out there. Why can't I spend one tenth of that or one one hundredth of that on talent?

• 23:38Speaker 01

  Yeah, okay. So then if that's the case, I mean, the assumption there is you're scaling revenue solely on, I mean, virality essentially, or just like high level brand marketing by saying things like bioweapons, I guess. But, but to me, I mean, I guess I will say this with a grain of salt because they have shown they're able to scale revenue already with this philosophy and this strategy. But to me, does it keep going? Or at a certain point, especially at that scale, are you going to start running into the Googles of the world or, or the hyperscalers? More, more regularly versus you, every individual software engineer is ready to pay you 200 bucks

• 24:21Speaker 01

  a month on their bank, on their work credit card versus that's a whole other level of competition.

• 24:29Alex Kantrowitz?

  Well, you would imagine it would be harder to 10x every time you 10x just because the law of large numbers comes into play, like going from one billion.

• 24:38Speaker 01

  Unless you're Nvidia.

• 24:39Alex Kantrowitz?

  Unless you're Nvidia, in which case that law doesn't really apply. You just substitute Jevon's paradox or whatever fun made up economic term you want, and you just keep printing money. I don't know. Getting from 10 to 100 billion is much, much harder.

• 24:52Speaker 01

  If they get to 10, I'm, I will tip my hat to Dario and team.

• 24:57Alex Kantrowitz?

  Yes, I will too. And speaking of 10, OpenAI is expected to surpass that this year, actually. So OpenAI hits 12 billion in annualized revenue, breaks 700 million ChatGPT, weekly active users. This is from Bloomberg. They roughly double their revenue in the first seven months of the year, reaching 12 billion in annualized revenue, says OpenAI. And I think there's this big stat, this is a big stat, ChatGPT has 700 million weekly active users. I mean, for a product to come on the market in November 2022, and potentially be on the road to a billion users by, let's say, mid 2026, that is insane from a company with no footprint beforehand.

• 25:41Speaker 01

  Yeah, to me, the most impressive part of that is there's no entrenched virality in the product. Like, yes, there's share your conversation, which actually I saw this post today and it was really interesting that apparently Google is now indexing shared chat GPT conversations and they're searchable. Conversation for another day, perhaps, but

• 26:04Alex Kantrowitz?

  No, let's talk about that now, actually, because I found that to be kind of weird and concerning that, like, I mean, maybe, because there's also been issues with like Meta's conversations showing up and you can like, there's almost nothing weirder than like, getting a chance to access people's private conversations or seemingly thinking private conversations with these bots. They, people will say the weirdest things to these chatbots, talking about their problems, their marriage, and the fact that like, you can, you can, you know, I guess, when you hit that share button, is there any clear messaging that this is like sharing to the entire, I mean, you should, you should intuitively know that you're going to share this to the entire internet, because if you drop it to somebody else, it's going to show up. But like, the idea that this could end up being

• 26:49Alex Kantrowitz?

  blasted to, you know, indexable Google is kind of crazy to me.

• 26:54Speaker 01

  Well, there's two points on that. One, it also just kind of brings me back to, I love the brawl. As, as, you know, like, Google being like, let's index that stuff, like, realizing, oh, well, we'll take some of their information right now and make it available and interesting through our own platform. But I think, yeah, the other part, so, so again, like, if you create a public share URL, it gets indexed. If you're like, sharing the conversation by itself and not creating that URL, it does not, same with Meta. There was like a big brouhaha like a month ago, and it was terrifying. Like, I was going through Meta AI and seeing some really weird conversations that were being publicly posted

• 27:38Speaker 01

  in feed. But again, people had, Meta even more so, it felt like accidentally hit share. But I think But I think so two parts of that there's getting to 700 million where virality is not entrenched is incredibly impressive. But this is where still how this stuff lives in our life, especially at the individual and consumer level. I think there's so much that's going to happen and so much that needs to be figured out at the business and enterprise level even more so. I think these things are a constant reminder from OpenAI that like ChatGPT suddenly imagine

• 28:18Speaker 01

  here if your conversations that your employees are sharing are suddenly indexed. But even remember, there's a story a few weeks ago where I forget which company was IPOing and someone asked for like internal financial documents and ChatGPT made up this entire, you know, like I have these documents, I got them from an investor who was part of the like testing waters discussions. And then, you know, the company denied it that they were real. So seemingly was hallucinated, but what's getting indexed where what's getting trained on where what's going to show up where? I think that there's there's definitely at the speed we're moving, a lot that's going to need to be figured out.

• 29:02Alex Kantrowitz?

  Okay, so this brings me, we talked a lot about GPT five last week, but I'm actually, I haven't gotten your perspective on it yet since you were out. It seems like GPT five is gonna come pretty soon. My guess is, I don't know, sooner rather than later. So what do you think about GPT five? You're obviously the product guy, not the model guy. Are you excited at all here?

• 29:26Speaker 01

  No, no.

• 29:28Alex Kantrowitz?

  You're the only person who who has been focused on AI who doesn't have any bit of anticipation for this new model.

• 29:34Speaker 01

  I don't know what step change to expect or see. The last few model releases have felt like iPhone releases for me. So I think, and again, this is going back to team product over team model. But do you think GPT five is going to be? Our AGI and what does that mean to you?

• 29:53Alex Kantrowitz?

  Well, last week on the show, I said, I think that once they release this, Sam Altman is going to call it AGI.

• 29:59Speaker 01

  Of course.

• 29:59Alex Kantrowitz?

  I've since, I mean, maybe it happens, maybe it does, but I've since sort of taken a step back and thought about like, what could be the downsides here? Because it is going to be this very big swing that they've been waiting to take and waiting to release for a long time. And Sam has also inflated the expectations, I would say, quite dramatically. So clearly they think they have a better model that they're going to hand over, which is, which is, I think, interesting and would be cool to see. But the other side is like, maybe the expectations are too high and people will sort of put this up as like a test for whether they're following the scaling laws and whether the improvement is as exponential or even linear. That, you know, in the way that they've been talking about.

• 30:43Alex Kantrowitz?

  And I read a post this week by Ryan Greenblatt, who is a researcher, an AI researcher, who, who asked, should we update against seeing relatively fast AI progress in 2025 and 2026? It's a very simple post. He's basically saying GPT-5 is expected to be released within three weeks. And rumors focused, that is, is substantially more focused on practical agentic software engineering. My expectation is that GPT-5 will be a decent amount better than O3 on agentic software engineering, but won't be substantially above trend. So we could end up getting a situation where we see a better model. It seems clear that this model is going to focus on an economically valuable area, which is coding.

• 31:25Alex Kantrowitz?

  But if it doesn't continue the trend, I think we're going to see a secondary news cycle here, which is that

• 31:33Alex Kantrowitz?

  people will be like, well, you know, we were promised AGI. It is not following the trend. It is showing those diminishing returns. And therefore we're going to, you know, go all of a sudden into skeptic mode for the rest of 2025 and 2026, that things will dramatically improve from here. And so I'm curious what you think about that.

• 31:55Speaker 01

  Yeah, I think it's it's gonna be like kind of a seminal moment. There's been something around the like allure of GPT-5 because it's been dangled in cryptic tweets for like code names for so long by Sam Altman that they have to deliver something magical. And agentic software engineering step up versus step change and also seeing the word "software engineering" there again, isn't as interesting to me like, is this another Sora or is this GPT-3? I don't know, like, I still am trying to envision, because they've never given us an idea or a picture of what it will be.

• 32:40Speaker 01

  And I get it, that that helps create the hype, but by not kind of clearly defining that direction for everyone and just trying to make it this kind of like mystical thing, I think it inflates expectations too much and I think it can certainly lead to skeptic mode.

• 32:57Alex Kantrowitz?

  So here's my prediction, there's gonna be an overreaction and maybe two overreactions. I'm almost 100% sure that we'll see at least one and maybe two overreactions to whatever happens.

• 33:10Speaker 01

  Wait, what are the two overreactions?

• 33:12Alex Kantrowitz?

  The first one is, the first one is while we reached ASI or AGI and the second one is like, oh, this does not actually continue the trend.

• 33:19Speaker 01

  Do you think he's gonna say ASI?

• 33:21Alex Kantrowitz?

  No, definitely not. Too early for that. I'd still, I still would be surprised but not shocked if they say it's AGI, but I think superintelligence is not gonna be.

• 33:32Speaker 01

  I don't know, just to, just to go after Ilya a bit, he just made his whole new startup Safe Superintelligence.

• 33:39Alex Kantrowitz?

  Be like, well, we've released Unsafe Superintelligence. Try to catch up.

• 33:41Speaker 01

  USI. And we're okay with it. USI, that's the future.

• 33:47Alex Kantrowitz?

  It's gonna be very interesting and for the nuanced non-overreactions, please continue listening to us because we'll do our best to break it down with the proper context and understanding. Well, one thing I think may not be overhyped because there's almost little hype about it is AI video and the rapid capabilities that are advancing in that area. So let's talk about that and whether we're in a market bubble right after this. And we're back here on Big Technology Podcast Friday edition with Ron John Roy of Margins, talking about all things AI this week. And Ron John, one of the interesting things that happened this week is you and I both dropped pretty fun stories about actual progress with AI video or dreams of AI video in our doc independently. So

• 34:31Alex Kantrowitz?

  why don't you lead with yours and then I'll tell you mine?

• 34:34Speaker 01

  All right. So the first thing I had come across, and again, what you had just mentioned, the lack of AI hype in this space, I think really is fascinating. Fable Studio, an Emmy-winning startup, just raised money, including from Amazon. They have an AI platform called Showrunner. So the idea here is you pay $10 to $40 per month, and you get to create your own animated shows or build on others' existing IP. And you can do things like insert yourself as a character into the show, or actually define the entire direction of the show. I think this is incredibly fascinating, because even for my son, like two years ago, or a year and a half ago, or whenever ChatGPT released Voice Mode, we

• 35:18Speaker 01

  started doing stories. And I would be like, you know, inserting him into it and telling me a story about a magical land and dragons and use the name John Jay. And he loved it. And I like, this whole world of kind of created entertainment and what kind of opportunities and experiences it'll open. I think this is where you'll see how to use, like, you know, it's one thing to just use ChatGPT and become slightly more productive in life, but to actually create these kinds of things, I think, is incredibly fascinating. But the other interesting part of this, though, is Fable is pitching studios to put their IP

• 35:59Speaker 01

  on its own platform. And even in the article that was on Business Insider, you know, they claimed that Disney might be interested. And like, the idea that, you know, will they ever give you, certainly Disney, I cannot imagine would be. But, but you know, like, imagine having access to making my own little South Park character and then making a story around it. It seems like that kind of creative energy, that's the fun stuff that will really push this technology to the mainstream, or just create things that, you know, would have been unimaginable before. Do you think this is real?

• 36:33Alex Kantrowitz?

  Yes.

• 36:33Speaker 01

  And do you think it's gonna work in this model?

• 36:37Alex Kantrowitz?

  Yes, but I will make a prediction here. And that is that this is, people are gonna say this is gonna displace Hollywood. And I think this is just an entirely new category in and of itself. Because I think one of the things, one of the great enjoyments of watching good entertainment is not being too involved in the creation process, sitting back and being surprised by the brilliance of the people that are creating the content that you're watching. Same with video games, like part of the enjoyment of video games is, you know, not necessarily like creating your own path, is just like following the, you know, brilliant output of thousands of game designers, if you're playing something like Assassin's Creed,

• 37:20Alex Kantrowitz?

  and not having to really worry about doing that yourself. Like, I think our brains have two different modes. We have like kind of doing and consuming, so to speak. And this type of AI creation actually puts the consuming bucket in a more of a creation place, if that makes sense. So I think that if we're gonna basically transpose the entertainment side of our lives into the creation part of our lives, it doesn't displace entertainment, it just creates a new category that sits maybe right between those two.

• 37:56Speaker 01

  Okay, I'm in, I buy that, I think, because if you think about like the entire creator economy, even the idea in like 20 years ago, that you could essentially create your own two to three minute TV show that would get distributed to millions of people was unimaginable. And now every single reel on your Instagram feed is that. And that's like someone giving access, given tools, given distribution to do that. And so you're right, but it hasn't displaced movies. All of us watch more like high production Netflix than ever before. So yeah, I think that's a fair idea that.

• 38:36Speaker 01

  It it's a different category. Maybe to me, it's almost the more it'll displace like creators and like, or it's a new category creator and it'll compete with kind of like influencer type reels versus it's it's not gonna ruin Avengers 26, but.

• 38:57Alex Kantrowitz?

  No, it's very interesting the way that you're thinking about this because the way that I came at this is like, are you gonna wanna create your own personalized stuff and watch that? As opposed to, do you wanna take these tools and create derivative work? That is viewed by lots of people. And that is actually, I think, more of a threat to the traditional entertainment industry than the first use case. Because it is something that's created by someone else, but viewed by viewed by many. The one thing I'll say is creating content is like really hard. And you might be able to entertain yourself on these platforms for a while, but it's much more difficult to, again, like have that storytelling capability. Even if you could like legitimately just prompt a Disney movie.

• 39:42Alex Kantrowitz?

  To have that story capability, telling capability that will sort of engage so many people. Now, maybe you wrote an amazing prompt and the AI is good enough that it can, you know, have a, it can be like a one line prompt that turns into a blockbuster hit. But I'm less bullish that that is something that we're gonna see anytime soon.

• 40:01Speaker 01

  Well, I'll add the caveat, creating good content is incredibly hard. In reality, creating content has gotten easier and easier and easier that it's like harder to find that good content. But, okay, I think I am now kind of after talking about this. Maybe this lives in that, like you can get personalized books or engraved or monogrammed like gifts, or maybe it lives in that realm more so than having any other major impact. But I still, I don't know, do you do? Do you know what's nuts? In 2009? Electronic Arts had a feature for FIFA where you could take a photo of yourself from like five different angles and upload it and create a player that actually was on

• 40:44Speaker 01

  the field. And I still, actually, this was one of my first blog posts ever, um, back in May 2009. Deep cut, yeah. No, but I literally, it was magical. I was like on the field with Fernando Torres and Liverpool and like scored a goal and was like hugging the team and these like pixelated faces in the background are, and it's crazy to me that that never took off. So clearly I was in the niche and not everyone was asking for that, but I still remember that. And that's why, like, there has to be some things that are just really fun and cool and make people excited to kind of experiment and do stuff. Like, yeah, in the video game, imagine you are the player, you see yourself in there. That's, that's kind

• 41:29Speaker 01

  of fun.

• 41:31Alex Kantrowitz?

  Let me give you a crazy stat. So streaming over the past year has been up 6%. Netflix has not increased share despite the fact that streaming as a, as a percentage of all TV or screen viewing has gone up. And the thing that is growing is YouTube. So that's maybe a counter argument to my earlier statement.

• 41:50Speaker 01

  Yeah, yeah, yeah. Yeah, I think even more so. Yeah. You don't have to, still creating good content that gets discovered and surfaced is, it's probably harder than ever if you're actually, if that's the model. But I don't know, I'm actually, I am excited about this. I think like, this is just going to be fun. This is, and maybe it lives in the world of you're buying some credits and sending it to your friends and like doing stupid stuff and, or funny stuff or stuff with your family and it's emotional, but I still think there's something here that's going to really define entertainment five years from now in ways that we didn't expect.

• 42:28Alex Kantrowitz?

  You might be right, and I'll just share my cool video AI video story of the week. This is from TechCrunch. Google's Notebook LM rolls out video overviews. Google announced on Tuesday that it's rolling out video overviews to Notebook LM. It's AI-based note-taking and research assistant. With the new capability, Notebook LM is taking more of a visual approach to helping users understand different topics and the ideas. The feature creates new visuals while pulling in diagrams, images, quotes, and numbers from uploaded documents to explain the content. Google says the feature is good for explaining data, demonstrating processes, and making abstract concepts easier to understand. I don't think people understand how freaking crazy this product is.

• 43:11Alex Kantrowitz?

  You legitimately upload documents, articles, whatever it is, and it will make you a video filled with like little animations and visuals that breaks down the topic you're trying to understand and like will like show it to you visually. Just the amount of, it just doesn't even seem like a real product. It's crazy to me.

• 43:33Speaker 01

  Well, and remember, Notebook LM kind of went viral for the first time with basically making a podcast out of the content, the text content that you feed it. And what was so, like, I'm just gonna say incredible about it is, is like the inflections of, and we say this as two guys talking on a podcast to each other. Like, like it almost made a mockery of us and everyone else podcasting because it, it really felt like a podcast and two people talking to each other and their inflection and their, the pauses and the kind of language that they were using. And so that's what went viral there. So I'm very interested in what it's going to do on the video side.

• 44:17Speaker 01

  But to me, one of the more interesting questions, I'm really curious what you think about is, is like I saw Joe Weisenthal, I saw a tweet around, it was like, the general trend is clear that all forms of passive media consumption are dying. Everything is some kind of conversation. With Notebook LM, you don't read a PDF. You create an AI-generated podcast to listen to. You ask questions. He's saying oral culture accelerates. Like, do you think, actually going back to our previous conversation, is passive media, I'm guessing you're gonna say not dead, but so much of... Instagram, I guess, has become more passive. Twitter is very interactive. Like, where do you see this going?

• 45:01Alex Kantrowitz?

  I think Joe is so wrong on this take for the reason that I said earlier.

• 45:05Speaker 01

  Hear that, Joe?

• 45:06Alex Kantrowitz?

  Joe, if you're listening, totally wrong take here. Why do people go to entertainment? It's escapism in a way. It's like, I want to just like, take, leave my world for a moment, and go into this crazy world that I would never be able to visit, except the fact that like, Netflix has made this game about, like, you know, this video about the squid game, and I want to watch that. And that you transport yourself there. You imagine yourself as the character as you watch, you're entertained, you get like adrenaline boosts by watching these things, maybe. So yeah, I don't, I don't think people are gonna want to let that go.

• 45:39Speaker 01

  No, but, but see, this is actually, when I watch now, I go to Reddit after and read threads. I like ask Google or ChatGPT in real time, who are the characters, who are the actors and what else have they been in? You know, like, it's no longer, unless I'm in the theater, completely passive. There's always kind of like ancillary actions or activities taking place around the content, which is good, which kind of like deep, more deeply immerses you in it, at least for me. Maybe I have attention span problems at home, but I think that's a common behavior.

• 46:14Alex Kantrowitz?

  I think you're right, and I think this is why our society is unhappy. Okay.

• 46:21Alex Kantrowitz?

  Let's talk about, let's talk about the market. We are in the thick of big tech earnings. We have about five minutes left, but I definitely want to touch on it. Bloomberg, the S and P 500 is a touch away from triggering sell signal, Bank of America says. This is from Bank of America's Michael Hartnett. Overbought markets can stay overbought as greed is harder to conquer than fear. He says that eighty, eighty-eight percent of the country's stock indexes are above their fifty to two hundred day moving averages. Oh, when, when they are eighty-eight percent above these moving averages, it's time to sell. Currently, eighty-two percent of them are above their moving averages, which is of course this technical behavior or not technical tracking that they do of the markets, but obviously like.

• 47:07Alex Kantrowitz?

  It would seem like we have a lot of froth. We have Nvidia at 4 trillion. We have Microsoft, which just became the second $4 trillion company, which is crazy. But then again, you do have earnings from these companies. And everybody's beating earnings expectations by a lot this quarter. What do you think this all means, Ron John?

• 47:26Speaker 01

  These, everyone is Nvidia now. Like, I mean, and as we speak, Facebook is, Meta is up 11% pre-market. Microsoft is up 9% and these are multi-trillion dollar companies. But Microsoft's Q4 revenue was up 18%. It's 76 billion, 18%. Like, I mean, having been in the business world a long time, like hitting near 20% growth when you're a startup or you're like a mid-level, mid-scale company is hard. It's hard as hell. So like, to do that at that scale is shocking. And I mean, it makes it not unreasonable that,

• 48:10Speaker 01

  and when your net income and your profit margin is 24%, 27 billion. Like, these numbers are nearly unfathomable. The same way I would feel when, uh, when you would see those Nvidia earnings about a year ago. So, things are, things are humming. I mean, there's no taking that away. Do you, do you see this continuing?

• 48:31Alex Kantrowitz?

  Well, every time I've said, I don't know if they can continue to grow. They've continued to grow. And it looks like the tariff threat is not going to be anywhere near as bad. And remember, the economy was humming coming into that. So, I think that it will continue now. It's funny because like I read-

• 48:49Speaker 01

  That was the death blow right there.

• 48:51Alex Kantrowitz?

  Right.

• 48:51Speaker 01

  Death blow.

• 48:52Alex Kantrowitz?

  Right. I read that. But the thing is that I read that headline and I was like, did the people in 2000 know that the, you know, massive growth that they were seeing was the beginning of a, you know, a crash, a bubble and a crash? Probably not. But then again, the earnings are there, like the money is coming in. And it is like the entire economy basically spending money on these services to try to get better at what they do. Right? Cloud revenue at Microsoft, I think you might have mentioned this. 39% year over year.

• 49:23Alex Kantrowitz?

  Azure, 39% and the full year revenue topping 75 billion and Microsoft's cloud revenue for the quarter, reaching 46.7 billion. It's real money. It's real money. So it seems like this can keep going, but maybe we're missing something.

• 49:44Speaker 01

  Those are the only words to put it. I'm near speechless. It seems like it can keep going. Maybe we're missing something. What could we be missing? I'm curious if others have thoughts, but it's a, they're good numbers. They're good numbers and they're not slowing down. Again, Meta's well up 22% on top line, up 36% on the bottom line. Yeah, these are, these are good numbers for any company at any scale, much less a multi-trillion dollar company.

• 50:13Alex Kantrowitz?

  Yeah. And capital expenditures are going up across the board. Big Tech is expected to spend 330 billion this year building largely data centers.

• 50:21Speaker 01

  Got to spend money to make money, right? If you're good at business.

• 50:25Alex Kantrowitz?

  If you're good at business, you make money. Get a business, you make money. Folks, if you, you know, I've heard, we don't learn things on big technology podcasts. We've just, you know, given one of the best business lessons, I would say, in the history of the show. Get a business, you're going to make money.

• 50:41Speaker 01

  I think that's the only way to end right there.

• 50:43Alex Kantrowitz?

  Put this thing on master class. Put it in the books. All right. It's worth the subscription. Ron, John, it's great to see you as always. Thanks for coming on the show.

• 50:51Speaker 01

  All right. See you next week.

• 50:52Alex Kantrowitz?

  All right. See you next week. Thank you, everybody for listening. Thanks again to all of our new and existing listeners and viewers. It's always great to be on this journey with you and thanks again for listening. On Wednesday, Replit CEO Amjad Masad is coming in to talk all about vibe coding. So I hope you'll stick with us for that. And also MG Siegler will be here on Monday for his monthly Monday spot to talk about Mark Zuckerberg's personal super intelligence. Thank you, Ron John. Thank you, everybody for listening and we'll see you next time on Big Technology Podcast.

• 51:24Speaker 02

  Now, more than ever, we need someone on our side. But Rob Whitman? Whitman's votes drove up costs, cut Medicaid, and risked 350,000 Virginians' health care to pay for tax breaks for himself and his donors. He's not on our side. Shannon Taylor will fight corruption, cut taxes, and bring down costs. This November, vote for Shannon Taylor. She's for us. Paid for by HMP, the HouseMajorityPac.com. Not authorized by any candidate or candidate's committee.

• 51:53Speaker 03

  In this life, you can't win.

• 51:55Speaker 04

  When the world gets you down.

• 51:56Speaker 03

  No!

• 51:57Speaker 04

  Escape to Pluto TV.

• 51:58Speaker 03

  This is objectively fantastic.

• 52:00Speaker 04

  Relax into the movies and shows you already know and love, all for free.

• 52:04Speaker 03

  Schwing!

• 52:05Speaker 04

  Even our news is uplifting.

• 52:06Speaker 01

  You stay classy, San Diego.

• 52:08Speaker 04

  Watch School of Rock, Crazy Ex Girlfriend, Anchorman: The Legend of Ron Burgundy, The X-Files, Wayne's World, and so much more.

• 52:14Speaker 03

  We're not worthy!

• 52:15Speaker 04

  The best things on TV really are free. You just have to know where to look. Escape to Pluto TV. Stream now, pay never.