DeepSeek’s release of a high-profile new AI model underscores a point we at OpenAI have been making for quite some time: the U.S. is in a competition with the Chinese Communist Party (CCP) that will determine whether democratic AI wins over the Chinese Communist Party’s (CCP) authoritarian version of the technology. The U.S. must come out on top–and the stakes could not be higher.
As President Donald Trump rightly said on Monday, DeepSeek “should be a wake-up call for our industries that we need to be laser-focused on competing to win.”
We couldn’t agree more.
For those familiar with AI and how the models get built, it wasn’t a huge surprise that someone was able to replicate parts of OpenAI’s o1 model several months after it was released. What’s notable is that it was not an American one that did so.
THE DEEPSEEK AI CHATBOT BURST ONTO THE SCENE: ARE FEARS ABOUT IT OVERBLOWN?
CCP leader Xi Jinping has made clear China wants to be the dominant player in AI by 2030, and the country is plowing enormous amounts of money into the AI infrastructure to compete with the U.S. They’re giving developers unfettered access to data; building enormous amounts of energy (ten nuclear facilities came online last year with another ten on course for this year); and are seeking to develop their own chip-manufacturing capabilities.
Though I suspect we will learn more about DeepSeek’s work that may ultimately impact how we understand the state of their technology (experts are already noticing that their system is slower to respond to user requests than other models), how new it actually is (some users have pointed out that the DeepSeek model says that it’s ChatGPT when asked), and what it cost (projects in of authoritarian countries have a tendency to be opaque), this weekend’s news shows that the CCP is all-in on the AI competition.
WHAT IS CHINESE AI APP DEEPSEEK?
It’s important to understand that the stock market panic over the release of DeepSeek’s R1 model is missing the big picture: when AI systems grow more efficient, we need more of the infrastructure that powers those tools, not less.
Think of it this way: in post-WWII America, just because companies in the U.S. and abroad designed more affordable cars over time did not mean that we stopped building highways. If anything, it made highways more valuable because people can travel farther and faster than they could before. In fact, the U.S. went big and built out the interstate highway system.
The same holds true with AI. More efficient models make computers more valuable than ever because we can achieve significantly greater outcomes with the same hardware, as President Trump noted in his remarks about DeepSeek on Monday.
Scaling up our AI infrastructure will scale up our AI capabilities, powering bigger breakthroughs than would have been imaginable even a few months ago in everything from healthcare and biotech to energy and national security.
Moreover, the most pressing issue in AI is the push towards AGI and superintelligence, which is the evolution of the technology to a point where it is able to help humans solve problems in science, health, and education that just a short time ago would have seemed like a miracle. This is the Super Bowl of AI, and the U.S. has to lead if we want to maximize the technology’s economic gains and ensure the world’s AI is built on democratic values.
The road to AGI and superintelligence requires investing in supercomputers, data centers, advanced chips, power generation facilities, transmission lines, human talent, and other key parts of the AI ecosystem. That’s the mission of the Stargate Project, the new venture that we unveiled at the White House last week that will immediately invest an initial $100 billion in new infrastructure, with activity already in the works on the ground in Texas. When it comes to accelerating the construction of democratic AI that can prevail over authoritarian AI, Stargate is a 21st century version of World War II’s Arsenal of Democracy.
Let’s make sure we understand the game being played, the nature of our competitors, and the stakes. If we do not accelerate our AI infrastructure build-out now, we’re effectively handing the future to the CCP.
The stakes are too high to let short-sighted market narratives or misinformation dictate our path. The game is on, and we need to play to win.
Chris Lehane is Chief Global Affairs Officer at OpenAI. Before joining OpenAI, he was Chief Strategy Officer and Operating Partner at Haun Ventures and previously led policy and communications at Airbnb from 2015 to 2021. He also co-founded a strategic consultancy, held various roles in the Clinton administration, and currently serves on Coinbase’s Board of Directors.
