President Donald Trump just announced Stargate, a $500 billion AI investment backed by SoftBank, Oracle, MGX, and OpenAI over the next four years.
What’s Actually Happening?
Let’s put it differently: OpenAI just secured a major customer—Stargate. SoftBank will finance the deal, while Oracle will provide the infrastructure for OpenAI’s AI services. This is good news for OpenAI. Training AI models is expensive, and open-source AI is catching up fast. OpenAI’s only path forward is securing exclusive access to data and users.As I wrote in The Race for AI Agents: “To dominate AI agents, companies need access to users and enterprise data.” — OpenAI and Oracle both lag behind in these areas. This deal sends a message.
The $500 billion Message To China
A $500 billion message: “We will catch up — we will outcompete you.” For President Trump, this message is aimed at China. But others heard it too, including Elon Musk, who missed the AI lead and has tried to catch up through lawsuits and ethical arguments. Not being part of Stargate, he quickly dismissed it as “half-baked.”
To which Sam Altman responded, “Wrong,” and then questioned:
“I realize what’s good for the country isn’t always good for your companies, but in your new role, I hope you’ll put America first.”
Easy for Altman to say—this deal is clearly great news for OpenAI. But is it equally good for America?
Why Announce This At The White House?
The AI race isn’t just about OpenAI vs. Musk’s Grok—it’s about who will dominate global technology and maintain hegemonic power. AI is set to reshape industries, and whoever leads in AI will reshape the world’s power structure. Back in 2019, I warned that China’s aggressive data collection strategy would give it a long-term AI advantage over the U.S. Today, the gap between the U.S. and China seems narrower than ever.
China’s DeepSeek’s R1
Take the latest quest for AI reasoning. I predicted in my podcast with Jasper Masemann that: “AI reasoning will be the main development in 2024.” By the end of 2024, Sam Altman claimed OpenAI now ‘knows’ how to build reasoning AI agents. How? I described the process in The Battle of Tech Giants: “o1 is designed for reasoning through an iterative, self-calling process.” Now, after less than 60 days, this technological edge is gone. China has caught up. DeepSeek’s R1, a Chinese AI model, uses the same self-critique and iterative learning approach — with stunning results.
AI Hegemony: Who Will Lead?
DeepSeek’s R1 is an open-source model, and soon many startup founders will experiment with it — indirectly helping China in its challenge to U.S. AI dominance. And the irony? This is happening despite U.S. efforts to block China from accessing AI technology — or perhaps because of those bans, which forced China to stay ahead by open-sourcing its models. Given all this, it’s no surprise that Trump chose the White House as the backdrop for a $500 billion “We will outcompete you” announcement.
Trump Rolls Back Biden’s AI Regulations
Alongside Stargate, Trump is also rolling back Biden’s Executive Order 14110, which imposed AI regulations. Those had been criticized by me and others for not being effective. As I wrote in Biden’s AI Plan: “…the White House seems to think AI can be controlled like nuclear weapons. But it’s not that simple.”
And clearly, it wasn’t. At least, the order didn’t stop OpenAI or China — if anything, it pushed them to innovate faster.
The AI Race Is Wide Open
With China pushing forward and the U.S. making massive investments, the AI race is far from settled. The core question remains the same: It’s not who has the best AI—but who controls the data and who builds products that businesses and people actually use. The journey is just beginning.
