The founder and CEO of Scale AI, and at just 25 years old he became the youngest self-made billionaire in the world. As we all know, ChatGPT has thrust AI into the global spotlight, positioning it as one of the most dominant trends in tech. This momentum is expected to continue through 2025, though the landscape will evolve as new technologies from China and other fast-growing regions begin to take shape. At the heart of this AI revolution is Scale AI. With a total of $1.35 billion raised at a current valuation of $14 billion, Scale AI is accelerating AI application development for major players like OpenAI, Meta, and Microsoft. To discuss what lies ahead for AI and its transformative potential, please join me in welcoming to the stage Felix Salmon from Axios in conversation with Scale AI founder and CEO Alexandr Wang. Can any nation truly win the AI race? Will China's AI leapfrog the West? Is AI the new global superpower? AI is our assistant, not the decision maker. We need to hear from the leaders and people that are shaping the AI landscape. My name is Alexandr Wang and I'm the founder and CEO of Scale AI.
Hello Doha, thanks for having us. And Alex, thanks for making it all the way here for this chat. It's going to be fascinating. We're talking in a period of geopolitical chaos, I think it's fair to say. And the stated title of this talk is like... Well, actually, you just ran an ad in the Washington Post the day before yesterday. You took out a full page ad in the Washington Post. What did it say?
Yeah, we took out this ad the day after the inauguration. So it said, 'Dear President Trump, we must win the AI war.' And that was basically it. Full page ad, those were the only words: 'America must win the AI war.'
Wow, it's kind of a little bit terrifying. Obviously you're an American, you went to the inauguration. I'm kind of fascinated. Quick, can we have a quick show of hands? How many people in this room would agree with Alex that America must win the AI war? Two. How many would disagree? Okay, this is not a friendly audience. This is an overwhelmingly skeptical audience when it comes to that. So make your case, because these people need to be convinced.
Well, first off, this is something I've been consistent about, dating back to fall of 2022, before even ChatGPT was released. I published a blog post about the AI war and how to win it from the perspective of an American. And effectively called out that it's clear that AI is going to fundamentally change the nature of national security, the nature of warfare. I think the degree to which this is true is somewhat underrated even today. And when there are these large discontinuous shifts, it's important that you acknowledge the moment. And we, as in America, that we were able to be leaders in this technological change. Other context: there are a number of papers and reports from the CCP, from the Chinese Communist Party and People's Liberation Army, dating back to 2017 and 2018, where they speak about how they believe that this new batch of technologies, AI being the central one, are an opportunity for the Chinese military to leapfrog current leaders. Despite the fact that they're currently behind, in their view they see these new technologies as a stepping stone or an opportunity, if they over-invest, to leapfrog Western powers. So that's kind of the context.
And just to be clear, one of the sort of fun facts about Scale AI, I don't know how familiar this audience is, you're mostly known for hiring a whole bunch of humans to train up AIs, but you're also a defense tech company.
Yeah, so we think about ourselves as the humanity-first AI company, and our goal is to ensure that we build AI models and applications that ultimately work for humanity. This breaks down into two core pieces. The first is that our data Foundry technology is used to power nearly every major AI model, from OpenAI to Google to Meta to xAI, and many of the leading large models. DeepSeek, not DeepSeek, I'm sure we'll talk about that in a bit. And then we also partner very closely with large enterprises and governments to ensure that they're able to apply the technology effectively. I think important context as well: I grew up in Los Alamos, New Mexico, with the birthplace of the atomic bomb, and both my parents were physicists who worked at the National Lab there. So my personal background is one where it's impossible to think about disruptive technologies as not critical to national security, or not fundamentally linked to how technology intersects with militaries, national security, and all the like.
Basically what you're saying is you are old enough to remember when Russia was the enemy.
I have no comment on that.
But no, I mean, in terms of this ad that you placed in the Washington Post and what you're talking about, the CCP and the PLA, it is clear that you see this as a war, that's the word you used in your ad, between China and the United States. And there is going to presumably, you want America to win this war. What is America fighting for in this war?
If you take a step back, it's very clear at the present moment there are two countries in which the preponderance of AI development and AI innovation are happening: the United States and China. And as DeepSeek demonstrated, it is a relatively close race. Obviously, in the United States, many of these innovations came from: the Transformer invented in the United States, GPT was invented in the United States, RLHF was invented in the United States, test-time compute was invented in the United States. It's very clearly the innovation hub. But Chinese technology has clearly caught up very quickly with models like DeepSeek, models from Alibaba and other firms. And so we're at a moment where I think you play the tape forward on AI, nearly every country in the world or nearly every company in the world will ultimately build AI technology on top of either the US technology stack or the Chinese technology stack. And so that's one of the most tangible fronts of the war, so to speak, or the most tangible area where you can see this AI war playing out: which technology stack becomes the sort of global standard for as much of the globe as possible. There have been multiple rounds of this. If you look back to the early days of the internet, American technology and American platforms became the global standards. But then you look towards what happened with the Belt and Road initiatives from the CCP, and you look towards the technology Huawei and other technologies that they exported, and you look today at internet technologies, TikTok and ByteDance are quite competitive with US firms. So which AI standard or which AI tech underpins the global order is the critical question.
And as we will remember from since the Snowden leaks came out, we know what that means. If you're building on an American technology stack, that basically means the Americans know everything you're doing. And if presumably conversely you're building on the Chinese technology stack, then that means the Chinese know what you're doing. That's the stake here. You're basically, if you're in the Middle East or in Europe or in Africa, you just need to sort of pick: do I want the Americans to have access to all of my secrets or the Chinese? And you're saying they should be the Americans.
No, I don't really see it as that being the key distinction.
So wait, the key distinction you just laid out was who owns the technology stack. If it's not a question of which government gets to have back doors into that technology stack, then what is it that makes that important? Why does it matter who owns the technology stack?
Well, I think first, one step back: as an American, 10 out of 10 times I'm going to advocate for the American stack as a proud American. But I think in addition to that, the critical question is what are the ideologies that are baked into the technologies? Ultimately, if you look at large language models or many of these AI technologies, there are a large number of technical decisions that have meaningful impact into how the model ultimately performs, or what is the information that the model ultimately gives, or what are the recommendations it might give to a person. So if you think about AI as a particularly important technology, which tech stack actually boils down to a question of what ideological decisions do you want to be made in the models. And I think probably the ones that matter the most to me personally are the question of free speech and the question of a democracy versus a communist society. And I think these are some of the key tensions that exist between the two systems.
