The Fight for AI Market Dominance | CNBC Marathon
China's latest AI breakthrough has leapfrogged the world. What took Google and OpenAI years and hundreds of millions of dollars to build DeepSeek says, took it just two months and less than $6 million. They have the best open source model, and all the American developers are building on that. Google and Character AI. Three of the biggest tech companies swallowing three of the biggest startups in the AI world without actually acquiring them. A playbook that Microsoft pioneered.
In March, it inked an unusual deal with inflection, a prominent AI startup with a chatbot to rival ChatGPT's. In June, Amazon and Adept: AI struck a deal. Amazon reportedly paid at least $330 million to license Adept Technology with $100 million retention bonus to employees who jumped ship. Bipedal humanoid robots using semantic intelligence, it's able to interpret commands from people and then make its own decisions about how to act. Robots like this one are catching the attention and billions of investment dollars from big tech companies like Amazon, Google, Nvidia and Microsoft. Elon Musk is betting the future of Tesla on these machines.
China's latest AI breakthrough has leapfrogged the world. I think we should take the development out of China very, very seriously. A game changing move that does not come from OpenAI, Google or Meta. There is a new model, that has all of the valley buzzing.
But from a Chinese lab called DeepSeek. It's opened a lot of eyes of like what is actually happening in AI in China. What took Google and OpenAI years and hundreds of millions of dollars to build, DeepSeek says, took it just two months and less than $6 million. They have the best open source model, and all the American developers are building on that. I'm Deirdre Bosa with the Tech Check Take: China's AI Breakthrough.
It was a technological leap that shocked Silicon Valley. A newly unveiled free, open source AI model that beat some of the most powerful ones on the market. But it wasn't a new launch from OpenAI or model announcement from Anthropic. This one was built in the East, by a Chinese research lab called DeepSeek, and the details behind its development stunned top AI researchers here in the U.S. First, the cost, the AI lab reportedly spent just $5.6 million to build DeepSeek version 3. Compare that to OpenAI, which is spending $5 billion a year, and Google, which expects capital expenditures in 2024 to soar to over $50 billion.
And then there's Microsoft that shelled out more than $13 billion dollars just to invest in OpenAI. But even more stunning how DeepSeek's scrappier model was able to outperform the lavishly funded American ones. To see the DeepSeek, new model.
It's super impressive in terms of both how they have really effectively done an open source model that does what is this inference time compute. And it's super compute efficient. It beat Meta's Llama, OpenAI's GPT-4o and Anthropic's Claude Sonnet 3.5 on accuracy on wide ranging tests. A subset of 500 math problems, an AI math evaluation, coding competitions, and a test of spotting and fixing bugs in code.
Quickly following that up with a new reasoning model called R1, which just as easily outperformed OpenAI's cutting edge, o1, in some of those third party tests. Today we released Humanity's Last Exam, which is a new evaluation or benchmark of AI models that we produced by getting, you know, math, physics, biology, chemistry professors to provide the hardest questions they could possibly imagine. DeepSeek, which is the leading Chinese AI lab, their model is actually the top performing, or roughly on par with the best American models.
They accomplished all that despite the strict semiconductor restrictions that the U.S . government has imposed on China, which has essentially shackled them out of computing power. Washington has drawn a hard line against China in the AI race, cutting the country off from receiving America's most powerful chips, like Nvidia's H100 GPUs. Those were once thought to be essential to building a competitive AI model, with startups and big tech firms alike scrambling to get their hands on any available. But DeepSeek turned that on its head, sidestepping the rules by using Nvidia's less performant H-800s, to build the latest model and showing that the chip export controls were not the chokehold DC intended. They were able to take whatever hardware they were trained on, but use it way more efficiently.
But just who's behind DeepSeek anyway, despite its breakthrough, very, very little is known about its lab and its founder, Liang Wenfeng. According to Chinese media reports, DeepSeek was born out of a Chinese hedge fund, called High Flyer Quant, that manages about $8 billion in assets. The mission on its developer site, it reads simply: "unravel the mystery of AGI with curiosity. Answer the essential question with long-termism." The leading American AI startups.
Meanwhile, OpenAI and Anthropic, they have detailed charters and constitutions that lay out their principles and their founding missions, like these sections on AI safety and responsibility. Despite several attempts to reach someone at DeepSeek, we never got a response. How did they actually assemble this talent? How did they assemble all the hardware? How did they assemble the data to do all this? We don't know. And it's never been publicized and hopefully we can learn that. But the mystery brings into sharp relief just how urgent and complex the AI face off against China has become.
Because it's not just DeepSeek. Other, more well-known Chinese AI models have carved out positions in the race with limited resources as well. Kai Fu Lee, he's one of the leading AI researchers in China, formerly leading Google's operations there. Now his startup "Zero One Dot AI," it's attracting attention, becoming a unicorn just eight months after founding and bringing in almost $14 million in revenue in 2024.
The thing that shocks my friends in the Silicon Valley is not just our performance, but that we train the model with only $3 million, and GPT four was trained by $80 to $100 million. Trained with just $3 million dollars. Alibaba's Qwen, meanwhile, cut costs by as much as 85% on its large language models in a bid to attract more developers and signaling that the race is on. China's breakthrough undermines the lead that our AI labs were once thought to have.
In early 2024, former Google CEO Eric Schmidt. He predicted China was 2 to 3 years behind the U.S . in AI. But now, Schmidt is singing a different tune. Here he is on ABC's "This Week." I used to think we were a couple of years ahead of China, but China has caught up in the last six months in a way that is remarkable. The fact of the matter is that a couple of the Chinese programs, one, for example, is called DeepSeek, looks like they've caught up. It raises major questions about just how wide open AI's moat really is.
Back when OpenAI released ChatGPT to the world in November of 2022, it was unprecedented and uncontested. Now, the company faces not only the international competition from Chinese models, but fierce domestic competition from Google's Gemini, Anthropic's Claude, and Meta's open source Llama Model. And now the game has changed. The widespread availability of powerful open source models allows developers to skip the demanding, capital intensive steps of building and training models themselves. Now they can build on top of existing models, making it significantly easier to jump to the frontier, that is the front of the race, with the smaller budget and a smaller team. In the last two weeks, AI research teams have really opened their eyes and have become way more ambitious on what's possible with a lot less capital.
So previously, you know, to get to the frontier, you would have to think about hundreds of millions of dollars of investment and perhaps $1 billion of investment. What DeepSeek is now done here in Silicon Valley is it's opened our eyes to what you can actually accomplish with 10, 15, 20, $30 million dollars. It also means any company like OpenAI that claims the frontier today, could lose it tomorrow. That's how DeepSeek was able to catch up so quickly. It started building on the existing frontier of AI, its approach focusing on iterating on existing technology rather than reinventing the wheel. They can take a really good big model and use a process called distillation. And what distillation is, is
basically you use a very large model to help your small model get smart at the thing that you want it to get smart at, and that's actually very cost efficient. It closed the gap by using available data sets, applying innovative tweaks, and leveraging existing models. So much so that deep Sik's model has run into an identity crisis.
It's convinced that its ChatGPT when you ask it directly, "what model are you?" DeepSeek responds, "I'm an AI language model created by OpenAI, specifically based on the GPT four architecture." Leading OpenAI CEO Sam Altman to post in a thinly veiled shot at DeepSeek just days after the model was released. It's relatively easy to copy something that, you know, works. It's extremely hard to do something new, risky and difficult when you don't know if it will work. But that's not exactly what DeepSeek did.
It emulated GPT by leveraging OpenAI's existing outputs and architecture principles while quietly introducing its own enhancements, really blurring the line between itself and ChatGPT. It all puts pressure on a closed source leader like OpenAI to justify its costlier model as more potentially nimbler competitors emerge. Everybody copies everybody in this field.
You can say Google did the transformer first. It's not. OpenAI and OpenAI just copied it. Google built the first large language models. They didn't productise it, but OpenAI did it into a productized way. So you can say all this in many ways. It doesn't matter.
So if everyone is copying one another. It raises the question, is massive spend on individual LMS even a good investment anymore? Now, no one has as much at stake as OpenAI. The startup raised over $6 billion in its last funding round alone, but the company has yet to turn a profit, and with its core business centered on building the models, it's much more exposed than companies like Google and Amazon, who have cloud and ad businesses bankrolling their spend. For OpenAI, reasoning will be key, a model that thinks before it generates a response. Going beyond pattern recognition to analyze, draw logical conclusions and solve really complex problems. For now, the startup's o1 reasoning
model. It's still cutting edge, but for how long? Researchers at Berkeley showed that they could build a reasoning model for $450 just last week. So you can actually create these models that do thinking for much, much less. You don't need those huge amounts of to pre-train the models, so I think the game is shifting.
It means that staying on top may require as much creativity as capital. DeepSeek's breakthrough also comes at a very tricky time for the AI darling. Just as OpenAI is moving to a for profit model and facing unprecedented brain drain. Can it raise more money at ever higher valuations if the game is changing? As Chamath Palihapitiya puts it...
Let me say the quiet part out loud: AI model building is a money trap. Those chip restrictions from the U.S . government, they were intended to slow down the race to keep American tech on American ground, to stay ahead in the race. What we want to do is we want to keep it in this country. China is a competitor and others are competitors.
So instead, the restrictions might have been just what China needed. Necessity is the mother of invention. Because they had to go figure out workarounds. They actually ended up building something a lot more efficient. It's really remarkable the amount of progress they've made with as little capital as it's taken them to make that progress.
It drove them to get creative, with huge implications. DeepSeek is an open source model, meaning that developers have full access and they can customize its weights or fine tune it to their liking. It's known that once open source is caught up or improved over closed source software, all developers migrate to that. But key is that it's also inexpensive. The lower the cost, the more attractive it is for developers to adopt. The bottom line is our inference cost is $0.10 per
million tokens, and that's 1/30th of what the typical comparable model charge. And where's it going? It's while the 10 cents would lead to building apps for much lower costs. So if you wanted to build a u.com or P erplexity or some
other app. You can either pay OpenAI $4.40 per million tokens, or if you have our model, it costs you just $0.10. It could mean that the prevailing model in global AI may be open source. As organizations and nations come around to the idea that collaboration and decentralization, those things can drive innovation faster and more efficiently than proprietary closed ecosystems. A cheaper, more efficient, widely adopted open source model from China that could lead to a major shift in dynamics. That's more dangerous because then they get to own the mindshare, the ecosystem.
In other words, the adoption of a Chinese open source model at scale that could undermine U.S . leadership while embedding China more deeply into the fabric of global tech infrastructure. There's always a good point where open source can stop being open source, too.
Right, so the licenses are very favorable today, but –it close it. –Exactly, over time, they can always change the license. So it's important that we actually have people here in America building.
And that's why Meta is so important. Another consequence of China's AI breakthrough is giving its Communist Party control of the narrative AI models built in China they're forced to adhere to a certain set of rules set by the state. They must embody "core socialist values." Studies have shown that models created by Tencent and Alibaba, they will censor historical events like Tiananmen Square, deny human rights abuse, and filter criticism of Chinese political leaders.
That contest is about whether we're going to have democratic AI informed by democratic values, built to serve democratic purposes, or we're going to end up with with autocratic AI. If developers really begin to adopt these models en masse because they're more efficient, that could have a serious ripple effect, trickle down to even consumer facing AI applications and influence how trustworthy those AI generated responses from chatbots really are. There's really only two countries right now in the world that can build this at scale, you know, and that is the U.S . and China, and so, you know, the consequences of the stakes in and around this are just enormous. Enormous stakes, enormous consequences and hanging in the balance: America's lead. For a topic so complex and new, we turn to an expert who's actually building in the space, and model agnostic.
Perplexity co-founder and CEO Arvind Srinivas, who you heard from throughout our piece. He sat down with me for more than 30 minutes to discuss deep seek and its implications, as well as Perplexitie's roadmap. We think it's worth listening to that whole conversation, so here it is.
So first I want to know what the stakes are. What like describe the AI race between China and the U.S . and what's at stake. Okay, so first of all, China has a lot of disadvantages in competing with the U.S.
Number one is, the fact that they don't get access to all the hardware that we have access to here. S o they're kind of working with lower end GPUs than us. I t's almost like working with the previous generation GPUs scrappily.
S o and the fact that the bigger models tend to be more smarter, naturally puts them at disadvantage. But the flip side of this is that necessity is the mother of invention. B ecause they had to go figure out workarounds, they actually ended up building something a lot more efficient. It's like saying, hey, look, you guys really got to get a top notch model, and I'm not going to give you resources, and then figure out something right? Unless it's impossible, unless it's mathematically possible to prove that it's impossible to do so, you can always try to come up with something more efficient, but that is likely to make them come up with a more efficient solution than America. And of course, they have open sourced it, so we can still adopt something like that here.
But that kind of talent they're building to do that will become an edge for them over time right? T he leading open source model in America's Meta's Llama family. It's really good. It's kind of like a model that you can run on your computer. B ut even though it got pretty close to GPT -4, and, at the time of its release, the model that was closest in quality was the giant 405B, not the 70B that you could run on your computer. And so there was still not a small, cheap, fast fashion open source model that rivaled the most powerful closed models from OpenAI and Anthropic. Nothing from America,
nothing from Mistral AI either. And then these guys just come out, with like a crazy model that's like 10x cheaper and API pricing than GPT-4 and 15x cheaper than Sonnet, I believe. R eally fast, 60 tokens per second, and pretty much equal or better in some benchmarks and worse in some others. But like roughly in that
ballpark of 4-O's quality. And they did it all with like approximately just 20, 48, 800 GPUs, which is actually equivalent to like somewhere around 1,500 or 1,000 1,500 H100 GPUs. That's like 20 to 30x lower than the amount of GPUs that GPT -4 is usually trained on, and roughly $5 million in total compute budget. They did it with so little money and such an amazing model, gave it away for free, wrote a technical paper, and definitely it makes us all question like, "okay, like if we have the equivalent of Doge for like model training, this is an example of that, right?" Right? Yeah. Efficiency is what you're getting at, so fraction of the price, fraction of the time.
Yeah. Dumbed down GPUs essentially. What was your surprise when you understood what they had done? So my surprise was that when I actually went through the technical paper, the amount of clever solutions they came up with, first of all, they trained a mixture of experts model. It's not that easy to train.
T here's a lot of, like, the main reason people find it difficult to catch up with OpenAI, especially on the MoE architecture, is that there's a lot of, irregular loss spikes. T he numerics are not stable so often, like, you got to restart the training checkpoint again, and a lot of infrastructure needs to be built for that. And they came up with very clever solutions to balance that without adding additional hacks. And they also figured out floating point-8 bit training, at least for some of the numerics.
And they cleverly figured out which has to be in higher precision, which has to be in lower precision. And to my knowledge, I think floating point training is not that well understood. I think most of the training in America is still running in FP16. Maybe OpenAI and some of the people are trying to explore that, but it's pretty difficult to get it right.
So because necessity is the mother of invention, because they don't have that much memory that many GPUs. They figured out a lot of numerical stability stuff that makes their training work. And they claimed in the paper that for majority of the training was stable, which means what? They can always rerun those training runs again and, on more data or better data. And then it only trained for 60 days. So that's pretty amazing.
Safe to say you are surprised. So I was definitely surprised. Usually the wisdom or, like I wouldn't say, wisdom, the myth, is that Chinese are just good at copying. So we start stop writing research papers in America. If we stop describing the details of our infrastructure or architecture and stop open sourcing, they're not going to be able to catch up. But the reality is, some of the details in DeepSeek v3 are so good that I wouldn't be surprised if Meta took a look at it and incorporated some of that in Llama Four–tried to copy them.
Right? I wouldn't necessarily say copy. It's all like, you know, sharing science, engineering. But the point is like it's changing. Like it's not like China is just copycat. They're also innovating.
We don't know exactly the data that it was trained on right? Even though it's open source, we know some of the ways and things that it was trained on, but not everything. And there's this idea that it was trained on public ChatGPT outputs, which would mean it just was copied. But you're saying it goes beyond that? There's real innovation. Yeah, look, I mean this they've trained it on 14.8
trillion tokens. The internet has so much ChatGPT. If you actually go to any LinkedIn post or X post now most of the comments are written by AI.
You can just see it like people are just trying to write. In fact, even within X there's like a Grok tweet enhancer, or in LinkedIn there's an AI enhancer or in Google Docs and Word. There are AI tools to like rewrite your stuff.
So if you do something there and copy paste it somewhere on the internet, it's naturally going to have some elements of a ChatGPT like training right? And there's a lot of people who don't even bother to strip away that I am a language model– right? –part. So, they just paste it somewhere and like it's very difficult to control for this. I think xAI has spoken about this too, so I wouldn't like disregard their technical accomplishment just because like for some prompts like who are you? Or like which model are you in response to that? It doesn't even matter in my opinion.
For a long time we thought, I don't know if he agreed with us. China was behind in AI. What does this do to that race? Can we say that China is catching up or has it caught up? I mean, like if we say that Meta is catching up to OpenAI and Anthropic, if you make that claim, then the same claim can be made for China catching up to America. A lot of papers from China that have tried to replicate o1, in fact, I saw more papers from China after o1 announcement that tried to replicate it, than from America. Like, and the amount of compute DeepSeek has access to is roughly similar to what PhD students in the U.S . have access to. By the way, this is not meant to like, criticize others. Like even for ourselves,
like, you know, I, for Perplexity. We decided not to train models because we thought it's a very expensive thing. Yeah. A nd we thought, like, there's no way to catch up with the rest. But will you incorporate DeepSeek into Perplexity? Oh, we already are beginning to use it.
I think they have an API, and we're also they have open source weights, so we can host it ourselves, too. And it's good to, like, try to start using that because it's actually, allows us to do a lot of things at lower cost. But what I'm kind of thinking is beyond that, which is like, okay, if these guys actually could train such a great model with us , you know, good team, like, and there's no excuse anymore for companies in the U.S., including ourselves, to like, not try to do something like that. You hear a lot in public from a lot of, you know, thought leaders in generative AI, both on the research side, on the entrepreneurial side, like Elon Musk and others say that China can't catch up. Like it's the stakes are too big.
The geopolitical stakes, whoever dominates AI is going to kind of dominate the economy, dominate the world. You know, it's been talked about in those massive terms. Are you worried about what China proved it was able to do? Firstly, I don't know if Elon ever said China can't catch up. I'm not aware of. Just the threat of China. He's only identified the threat of letting China, and you know, Sam Altman has said similar things. We can't let China win the race. You know, it's all I think you got to decouple what someone like Sam says to like what is in his self-interest.
Right? Look, I think the my point is, like, whatever you did to not let them catch up didn't even matter. They ended up catching up anyway. Necessity is the mother of invention. –Exactly–Like you said. And you it's actually, you know what's more dangerous than trying to do all the things to like, not let them catch up and, like, you know, all this stuff is what's more dangerous is they have the best open source model, and all the American developers are building on that. Right. That's more dangerous.
Because then, they get to own the mindshare, the ecosystem. If the entire American AI ecosystem, look, in general, it's known that once open source is caught up or improved over closed source software, all developers migrate to that. It's historically known, right? When Llama was being built and becoming more widely used, there was this question should we trust Zuckerberg? But now the question is should we trust China? That's a very–you should trust open source. That's the, like it's not about who is it Zuckerberg or is it? Does it matter then if it's Chinese, if it's open source? Look, it doesn't matter in the sense that you still have full control. Y ou run it as your own, like set of weights on your own computer. You are in charge of the model. But, it's not a great look for our own like talent to like, you know, rely on software built by others, even if it's open source.
There's always like a point where open source can stop being open source, too right? So the licenses are very favorable today. But if you close it. Exactly, over time, they can always change the license. So it's important that we actually have people here in America building. And that's why Meta is so important.
Like I look I still think Meta will build a better model than DeepSeek v3 and open source it, and what they call it Llama 4 or 3 point something, doesn't matter. But I think what is more key is that we don't like, try to, uh, focus all our energy on banning them and stopping them and just try to outcompete and win them. That's just that's just the American way of doing things. Just be better.
And it feels like there's, you know, we hear a lot more about these Chinese companies who are developing in a similar way, a lot more efficiently, a lot more cost effectively, right? Yeah. Again, like, look, it's hard to fake scarcity, right? If you raise $10 billion and your desire to spend 80% of it on a compute cluster, it's hard for you to come up with the exact same solution that someone with $5 million would do. And there's no point, no need to, like, sort of berate those who are putting more money in. They're trying to do it as fast as they can. When we say open source, there's so many different versions. Some people criticize meta for not publishing everything, and even DeepSeek itself isn't totally transparent. Yeah, you can go to the limits of open source and say, I should exactly be able to replicate your training run. But first of all,
how many people even have the resources to do that and compare? Like, like I think the amount of detail they've shared in the technical report, actually Meta did that too, by the way. Meta's Llama 3.3 technical report is incredibly detailed and very great for science. So the amount of details they get these people are sharing is already a lot more than what the other companies are doing right now. When you think about how much it costs DeepSeek to do this less than $6 million, think about what OpenAI has spent to develop GPT models.
What does that mean for the closed source model, ecosystem trajectory, momentum? What does it mean for OpenAI? I mean, it's very clear that we'll have like an open source version 4-o, or even better than that, and much cheaper than that, open source, like completely this year. Made by OpenAI? Probably not. Most likely not. And I don't think they care if it's not made by them. I think they've already moved to a new paradigm called the o1 family of models.
A nd I looked at I can't like Ilya Sutskever came and said, pre-training is a wall, right? So, I mean, he didn't exactly use the word, but he clearly said the age of pre-training is over. –Many people have said that. Right? So, that doesn't mean scaling has hit a wall. I think we're scaling on different dimensions now. The amount of time model spends thinking at test time, reinforcement learning, like trying to like make the model, okay, if it doesn't know what to do for a new prompt, it'll go and reason and collect data and interact with the world, use a bunch of tools. I think that's where things are headed, and I feel like OpenAI is more focused on that right now.
Yeah. I nstead of just the bigger, better model? Correct. Reasoning capacities, but didn't you say that DeepSeek is likely to turn their attention to reasoning? 100%, I think they will, and that's why I'm pretty excited about what they'll produce next. I guess that's then my question is, sort of, what's OpenAI's moat now? Well, I still think that, um, no one else has produced a system similar to the, o1 yet, exactly. I know that, like, there's debates about whether o1 is actually worth it.
Y ou know, on maybe a few prompts, it's really better. But like most of the times it's not producing any differentiated output from Sonnet. But, at least the results they showed in o3 where, they had like competitive coding performance and almost like an AI software engineer level. Isn't it just a matter of time, though, before the internet is filled with reasoning data, that DeepSeek? Again, it's possible, nobody knows yet.
Yeah. So until it's done, it's still uncertain. Right? Right. So, maybe that uncertainty is their moat. That, like, no one else has the same, reasoning capability yet, but will by end of this year, will there be multiple players even in the reasoning arena? I absolutely think so. So, are we seeing the commoditization of large language models? I think we will see a similar trajectory, just like how in pre-training and like post training, that sort of system, 4o getting commoditized, where this year will be a lot more commoditization there. I think the reasoning kind of models will go through a similar trajectory where in the beginning, 1 or 2 players really know how to do it, but over time– T hat's, and who knows, right? Because OpenAI could make another advancement to focus on.
But right now reasoning is their moat. By the way, if advancements keep happening again and again and again, like, I think the meaning of the word advancement also loses some of its value, right? Totally. Even now, it's very difficult. Right. Because there's pre-training advancements. Yeah. And then we've moved into a different phase. Yeah, so what is guaranteed to happen is whatever models exist today, that level of reasoning, that level of multimodal compute capability in like 5 to 10x cheaper models, open source, all that's going to happen. It's just a matter of time.
What is unclear is if something like a model that reasons at test time will be extremely cheap enough that like, we can just all run it on our phones. I think that's not clear to me yet. It feels like so much of the landscape has changed with what DeepSeek was able to prove.
Could you call it China's ChatGPT moment? Possible, I mean, I think it certainly probably gave them a lot of confidence that, like, you know, we're not really behind. No matter what you do to restrict our compute. Like, we can always figure out some workarounds. And yeah, I'm sure the team feels pumped about the like, you know, results.
How does this change, like the investment landscape, the hyperscalers that are spending tens of billions of dollars a year on CapEx, they've just ramped it up huge. And OpenAI and Anthropic that are raising billions of dollars for GPUs, essentially. But what DeepSeek told us is you don't need you don't necessarily need that.
Yeah. I mean, look, I think it's very clear that they're going to go even harder on reasoning because they understand that, like whatever they were building the previous two years is getting extremely cheap, that it doesn't make sense to go justify raising that. Spending proposition the same, do they need the same amount of, you know, high end GPUs, or can you reason using the lower end ones that DeepSeek has used? Again, it's hard to say no until proven it's not. But I guess like in the spirit of moving fast, you would want to use the high end chips and you would want to like move faster than your competitors. I think, like the best talent, still wants to work in the team that made it happen first. You know, there's always some glory to like, who did this actually? Like who's the real pioneer, versus who's the fast follower, right? That was like kind of like Sam Altman's tweet kind of veiled response to what DeeoSeek has been able to, h e kind of implied that they just copied and anyone can copy, right? Yeah, but then you can always say that, like, everybody copies everybody in this field.
You can say Google did the transformer first. It's not OpenAI and OpenAI just copied it. Google built the first large language models. They didn't productise it, but OpenAI did it into a productized way. So you can you can you can say all this
in many ways. It doesn't matter. I remember asking you being like, you know, why don't you want to build the model? Yeah. That's, you know, the glory, and a year later, just one year later, you look very, very smart to not engage in that extremely expensive race that has become so competitive. And you kind of have this lead now in what everyone wants to see now, which is like real world applications, killer applications of generative AI. Talk a little bit about like that decision and how that's sort of guided you, and where you see Perplexity going from here.
Look, one year ago, I don't even think we had something like, this is what, like 2024 beginning, right? I feel like we didn't even have something like a 3.5, right? Um, we had GPT -4, I believe, and it was kind of the, nobody else was able to catch up to it. Yeah. B ut there was no multimodal nothing. A nd my sense was like, okay, if people with way more resources and way more talent cannot catch up, it's very difficult to play that game. So let's play a different game. Anyway, people want to use these models. And there's one use case of asking questions and getting accurate answers with sources, with real time information.
Accurate information. T here's still a lot of work there to do outside the model, and making sure the product works reliably, keep scaling it up to usage. Keep building custom UIs. There's just a lot of work to do, and we will focus on that. And we would benefit from all the tailwinds of
models getting better and better. That's essentially what happened, in fact, I would say, S onnet 3.5 made our products so good, in the sense that if you use Sonnet 3.5 as the model choice within Perplexity,
it's very difficult to find a hallucination. I'm not saying it's impossible, but it dramatically reduced the rate of hallucinations, which meant the problem of question-answering, asking a question, getting an answer, doing a fact checks, research, going and asking anything out there because almost all the information is on the web, was such a big unlock, and that helped us grow 10x over the course of the year in terms of usage. And you've made huge strides in terms of users. And, you know, we hear on CNBC a lot like big investors who are huge fans. Yeah. Jensen Huang himself, right?
He mentioned it the other in his keynote the other night. He's a pretty regular user, actually, he's not just saying it. He's actually a pretty regular user. So, a year ago, we weren't even talking about monetization because you guys were just so new and you wanted to, you know, get yourselves out there and build some scale. But now you are looking at things like that, increasingly an ad model, right? Yeah. We're experimenting with it. I know there's some controversy on like, why should we do ads, whether you can have a truthful answer engine despite having ads and in my opinion, we've been pretty proactively thoughtful about it where we said, okay, as long as the answer is always accurate, unbiased and not corrupted by someone's like advertising budget, only you get to see some sponsored questions, and even the answers to those sponsored questions are not influenced by them.
And questions are also like, you know, not picked in a way where it's manipulative. Sure, there's some things that the advertiser also wants, which is they want you to know about their brand, and they want you to know the best parts of their brand, just like how you go and if you're introducing yourself to someone you want to, you want them to see the best parts of you, right? So that's all there. But you still don't have to click on a sponsored question, you can ignore it. And we are only charging them CPM right now.
So we're not we ourselves are not even incentivized to make you click yet. So I think considering all this, we're actually trying to get it right, long term. Instead of going the Google way of forcing you to click on links. I remember when people were talking about the commoditization of models a year ago and you thought, oh, it was controversial, but now it's not controversial. It's kind of like that's happening and you're keeping your eye on that is smart. By the way, we benefit a lot from model commoditization, except we also need to figure out something to offer to the paid users, like a more soft, skilled research agent that can do like multi-step reasoning. Go and like do like 15 minutes worth of
searching and give you like an analysis, an analyst type of answer. All that's going to come, all that's going to stay in the product, nothing's changed there. But there's a ton of questions every free user asks, day-to-day basis that that needs to be quick, fast answers like it shouldn't be slow, and all that will be free, whether you like it or not. It has to be free. That's what people are used to.
And that means like figuring out a way to make that free traffic also monetizable. So you're not trying to change user habits. But it's interesting because you are kind of trying to teach new habits to advertisers. They can't have everything that they have in a Google 10 blue links search. What's the response been from them so far? Are they willing to accept some of the trade offs? Yeah, I mean that's why they are trying stuff like Intuit is working with us. And then there's many other brands, Dell, like all these people are working with us to test right? They're also excited about hey, look, everyone knows that like whether you like it or not, 5 to 10 years from now, most people are going to be asking AIs most of the things and not on the traditional search engine.
Everybody understands that. Everybody wants to be early adopters of the new platforms, new UX and learn from it and build things together. Not like they're not viewing it as like, okay, you guys go figure out everything else and then we'll come later.
I'm smiling because it goes back perfectly to the point you made when you first sat down today, which is necessity is the mother of all invention, right? And that's what advertisers are essentially looking at. They're saying this field is changing. We have to learn to adapt with it. O kay. Arvind, I took up so much of your
time. Thank you so much, for taking the time. Hey, Digit, take the box with the heaviest animal and move it to tower four. This is Digit, a robot who sort of looks like a person, hence the name Humanoid Robot. Technically, these are called bipedal humanoid robots. Using semantic intelligence,
it's able to interpret commands from people and then make its own decisions about how to act. So the goal is for him to be able to interpret normal human language, to say, hey, I need you to pick this box up. Help me out in this instance, is that where you see this going? Yeah. And I think it will generally be probably less of a I need you to do this one thing for me and more of a do this for me forever in the corner of the facility.
Robots like this one are catching the attention and billions of investment dollars from big tech companies like Amazon, Google, Nvidia and Microsoft. Elon Musk is betting the future of Tesla on these machines. As you see Optimus develop, it's really going to transform the world. I think to a degree even greater than the cars. Some also argue robots like Digit can solve the world's labor crisis, filling jobs that are too dangerous or that people simply don't want. They may even replace an aging workforce as people around the world have fewer kids.
Today in manufacturing, we are short about 300,000 people, and it's something very similar in warehousing and logistics. So we're somewhere around 6 or 700,000 jobs we can't fill. The idea of robots isn't exactly new.
Here I am, sir. So why all of a sudden attention? The big driver, artificial intelligence. These bots have seen quantum leaps in what they're capable of in just the past few years, thanks to AI. Generative AI is really a key unlock overall for what you can get a robot to do, let alone a humanoid robot. Robotics is where AI meets reality.
We are really at the cusp of solving one of the grand challenges of humanity. It'll change labor forever. There is probably a need in the future once the humanoids get a bit more clever, a bit more perhaps autonomy. But until that point, I think the market will be fairly limited to PR spectacles. It all sounds great, but will we ever trust robots working in our houses, schools, and nursing homes? Will they ever be safe enough? And how lifelike is too lifelike? And how should the US think about global adversaries building similar fleets of humanoids? CNBC explores the rise of these AI driven humanoids. And if they're really a cure all for our global workforce problems, or if this is another tech bubble.
This hardware has been around for decades. Companies like Boston Dynamics, Honda, Sony and others have wowed the public with early versions of these robots. Why would we want humanoids? The prevailing sort of answer has been the world is built to be occupied by humans. We want robots that are versatile, that can do a wide range of things, and having it adopt the humanoid form factor always made a lot of sense. Recent leaps in artificial intelligence have resulted in leaps for robotics.
The data that they use to train these robots is based in real world scenarios. Now, a robot can be trained the same way a human is. We have this technology called Teleoperation. The person does the thing 200 times, we record all that data, and then we use that data to train these models. And the AI models are very similar to the GPT style generative AI models. You feed in the 200 trajectories, and the system learns how the task is being done, and then the robot will do the task autonomously.
If I go into a new space, I'm now not looking at spending months trying to code that problem. I can potentially just generate it straight out of GenAI and be able to have Digit interact with new objects and in new environments without having to develop at all. AI models require massive quantities of data to train off of, and this is no different. If you show the robot enough things, it starts to be able to do things that it hasn't been shown before. Big tech is very interested in the big potential this technology promises. If you're going to do AI at the frontier, you need to be partnered with Microsoft or Nvidia or Google or one of the big players.
There is no other way. They have resources that nobody else has, even governments. Nvidia has been a great partner up until this point. We're using everything from their hardware to their simulation, and then recently have started working with them on foundation models as well.
One of this industry's biggest proponents is Elon Musk. He's made some bold predictions that Tesla's robot, Optimus, could propel it to a $25 trillion market cap, and that it will amount to a majority of Tesla's long term value. With demand as high as 10 to 20 billion units. Tesla is arguably the world's biggest robotics company because our cars are like semi-sentient robots on wheels. And with the full self-driving computer and
all the neural nets, it kind of makes sense to put that onto a humanoid form. And it's intended to be friendly, of course, and navigate through a world built for humans and eliminate dangerous, repetitive and boring tasks. Musk isn't alone in believing humanoid robots could change the world. Investors are pouring millions into startups, with the market expected to grow to $38 billion by 2035.
The funding that has gone into some of these companies has been absolutely huge, and I think that kind of the parallel with the AI funding. This is the biggest market in the world. I mean, effectively, this has the potential to change the way we live and work pretty dramatically.
Humanoid robots are closer to being a real part of our workforce than you might think. Some companies are already deploying them in factories and warehouses. Tesla claims it has two Optimus humanoid robots in its factory. During Tesla's 2024 first quarter earnings call, Musk said he believes Optimus will be performing tasks in Tesla's factories by the end of the year, and that it could start selling the robot to outside customers by the end of 2025. At the company's 2024 investor day in June, Musk predicted it could have over a thousand or a few thousand robots working at Tesla next year.
Musk saying that Optimus will be bigger than the cars. I presume that means that they are spending huge amounts of this. I wouldn't underestimate them. Digit, created by Oregon based Agility Robotics, is helping Amazon in early stage testing at its Sumner, Washington, fulfillment center and innovation lab. We've been working with them recycling totes. The arms are capable of handling a wide variety of different payloads, up to about 33 -ish pounds. Agility says it plans to keep expanding the scope of work that Digit is capable of.
We envision an app store for robots out in the future where if you need Tote recycling app, you can go into the App Store and download that onto your robot. And according to the company, there is plenty of demand. Agility is building a factory in Salem, Oregon to keep up with orders. We call it robo fab, we will be online this summer, and in a few years, have a capacity of about 10,000 robots per year.
And where will those go for the most part? To many, many customers, but largely initially in the logistics warehousing space, the next big market we see is automotive, retail, and then eventually into markets like healthcare. Several other startups are developing similar humanoid robots. Sanctuary AI launched in 2018, in Vancouver, Canada, unveiled its latest robot last year.
Phoenix, a five foot seven robot capable of lifting up to 55 pounds. It looks a bit different from other humanoid designs, trading its legs for wheels. Robots with legs. The upper body, including the hands, have to be very weak and light. So instead of doing that,
we put our product on a wheeled base and because we made that trade off, we can build very powerful, very precise, very fast motors in the upper body. The company deployed early iterations of its robot with Canadian Tire, completing front and back of store tasks such as picking and packing merchandise. Robots were asked to do everything from greeting people when they come through the door to actually putting things on trucks. Sanctuary says it's close to releasing its eighth generation robot in the next few months, and has partnered with automotive manufacturer Magna to help build its robots at scale. Optronics started in 2016 as a spinout from the Human Centered Robotics Lab at the University of Texas at Austin. It began with an initial project to help NASA
build a generalized humanoid robot. The company says it's now on its eighth version of a humanoid. All of that has culminated in building the robot that we essentially always dreamed of building a robot called Apollo, which is a mass manufacturable commercial humanoid robot. With a max payload of 55 pounds and a swappable four hour battery. Apollo is designed to help support
logistics and manufacturing to start with retail as its next focus. We've got to prove out sort of the simpler tasks, but my dream is for Apollo, hopefully to be ready in time to help my parents, hopefully to help me as I get old. The company says it has deployed Apollo in pilots with Mercedes-Benz, GXO and others.
We've started with pretty simple tasks things like moving boxes or moving cases from one place to another. What we're moving into is doing more dexterity and more end to end tasks, and then hope to be in full commercial launch by the end of next year. It's no wonder tech companies have taken notice. Some Wall Street analysts predict these robots are the next must have device, not unlike a smartphones or EVs, but they also say such robots would be vital for manufacturing and dangerous work. But they would also help with elderly care and fill in labor shortages in factories.
There's already too few workers to fill all of the world's manufacturing jobs. It's an estimated shortage of 500,000 people, and by 2030, Goldman Sachs thinks that will grow to a shortage of 2 million workers. Imagine I could give you a labor force that costs the price of electricity, you know, a few cents an hour per worker. And they were equipped with the kinds of general intelligence that you might imagine comes like in science fiction. There's going to be things that you think of that will dramatically improve your life no matter who you are. We've only automated 10% of automotive manufacturing.
If we can automate a much higher degree, it would dramatically change the economy. We have the ability to add new tools. You can imagine if you needed to put a rivet into a car or something, you could actually have the screw produced in the hand. But robots taking human jobs can be a contentious topic. What would you tell those people who you know there are certain jobs that might be lost as a result of this? Actually, we've heard from some of those people, and what their jobs start to evolve into is the manager of the robot fleet.
They can be deployed in coal mines, in fires, for rescue efforts, where it might not be as safe for a human being. There are just some jobs that people don't want. Proponents say a humanoid can help fill those. There's about 10 million open jobs in the United States alone. We could build and maintain 10 million of the robots that we're talking about without touching a single job.
If you have to live sort of 55 pounds, somewhere between 50 and 100 times an hour. That's a serious workout. So we're taking these dirty, dull and dangerous jobs first.
How far away are we from Digit doing your laundry? That's probably more along the lines of, you know, a decade or more. Is there a kill switch? For lack of a better term. There is a kill switch. It's the big red button over there. The big red button over there. These are 140 pounds.
They have a lot of torque in their hands and arms and legs. And so you have to be very conscious of the situation when robots are interfacing with humans. The biggest sort of short term obstacle we have is safety. The UN has sort of put together a committee that's called AI for good, which is basically coming up with policy recommendations at a UN level for how should countries think about AI and also think about robotics. For Agility's, deployment of Digit its robots operate away from human workers for safety reasons. So right now digit is Non-collaborative and we're working to what's called collaborative safety, which is being able to formally verify, according to international standards, that Digit is safe to be in close proximity to a person.
We have very tight safety regulations today. And then eventually OSHA is the workplace safety organization that actually has to approve any deployment. So it's very regulated to make sure that we don't get bad accidents. Another big roadblock: the cost. Humanoid robots are expensive, complex pieces of machinery. For them to go mainstream, that's going to have to change.
It's really expensive to even try to do this, and you have to have investors that are willing to take risks because, you know, nobody's done this before. Elon Musk came out with this idea that it should cost less than $20,000, and that would be a level where mass adoption would be possible. I think we'll have to wait quite a few years, probably a decade at least.
Agility says its robots can be purchased up front, with an additional software as a service fee, or a bundled monthly fee for robots as a service. As long as they have the robot, we help maintain it, take care of it, keep the software updated and they just pay on a monthly basis. What does one of these robots cost? How should people think about it? Is it like buying a car or a boat? I'd say it's like buying an expensive car, but the costs are coming down very, very quickly. While the U.S. has seen activity in humanoid robots explode, China is giving the West a run for its money. It already dominates the industry, surpassing Japan in 2013 as the world's largest installer of industrial robots, and now accounts for more than half of the global total.
China, the market is absolutely huge. For example, in mobile robots, it's the biggest in the world. The only other company in the West that has anything similar to the quantities that the companies have in China is Amazon. But Chinese companies are catching up
fast. If you just count the number of humanoid robots that have come onto the scene over the last year. About half of them are from Chinese companies.
Interestingly, you've seen companies like BYD also invest. There's a broader effort to get the capital into this industry in China. One of China's biggest competitive advantages is cost.
They need to source these parts almost custom made. Being based in China, they're closer to the factories that can make these parts, potentially at a lower cost. Unitary in China. They came out with this $16,000 humanoid, which is a very exciting development.
But what can the robot actually do? That's a huge question. I think an interesting space to watch will be in the components. Once there are more standards and potentially the factories that have been making some of these components for the humanoid start ups in China, they could also sell those parts overseas. Competitiveness in this space is crucial for the U.S., which has been seeking to keep its edge over China as geopolitical tensions rise. Even though the U.S. invented the very first
robot, 100% of all major industrial robots were produced outside of the U.S. by foreign companies. I think of the humanoid race as effectively the next space race, and it has major implications for both national competitiveness and national security.
In terms of large economics, it is China versus the U.S. We need to make sure that the big applications for this are in the U.S. and in Europe, because if it's not, then we're going to lose that arms race. This is a real arms race and we better make sure we win it. There are still very real roadblocks.
Cost and safety are the biggest near-term issues. Finally, there's questions over how to regulate the space, similar to questions on how to regulate AI. Even the robot's biggest advocates say they want to put up guardrails and make sure those eventually working alongside humanoids aren't put in harm's way. But they're also competitive, and they want this industry to innovate with an ability to move quickly enough to keep up with rapid progress abroad. This is really the beginning of a new industry. Think of this like the personal computer in the early 80s. It will have potentially bigger impacts
than the personal computer. These robots are going to give us all back time. Startups are becoming pawns in the megacap game of chess. Microsoft absorbs Inflection. B ig organizational change happening at Microsoft around AI.
Amazon consumes adept. Its startup was valued at more than $1 billion. Now, most of the company is now going to Amazon. Google and Character.AI.
Three of the biggest tech companies swallowing three of the biggest startups in the AI world without actually acquiring
2025-05-28 09:37
