As Vice President JD Vance heads to Paris for the AI Action Summit, the Trump administration will be doubling down on the message that’s underpinned their first two weeks in office: “We’re not waiting for the rest of the world.”
Their approach to leadership is right on time because it’s clear our adversaries have been busy. Late last month, a small Chinese startup named DeepSeek stunned the AI world by releasing models that match the performance of industry leaders such as OpenAI and Anthropic. Even more surprising, they claim to have done this with far less computing power and without access to the latest American chips, due to export controls.
FULL LIST OF EXECUTIVE ORDERS, ACTIONS, AND PROCLAMATIONS TRUMP HAS MADE AS PRESIDENT
In light of this revelation, some political leaders such as former FTC Chair Lina Kahn, have declared American tech leadership dead. Others rushed to conclude that this proves AI progress doesn’t require massive computing resources. But they’re missing the real lesson: Innovation is shaped and constrained by the infrastructure we have.
While much remains to be verified, reports suggest DeepSeek is using Nvidia’s hardware in a highly specialized way – like rebuilding a car engine by hand instead of using the automatic transmission the car comes with. The CCP-backed company succeeded by taking a radically different approach to AI development because of the infrastructure they have — one that U.S. companies might never have discovered because of the walled AI garden so many build in now.
What this actually reveals is that there are different ways to build AI systems. But most U.S. companies are so deeply invested in Nvidia’s ecosystem — from hardware to proprietary software to developer tools – that exploring alternative approaches has become cost-prohibitive. And Nvidia actually markets this black-box approach as part of their appeal.
This is impressive but not game-changing on its face. DeepSeek faced the same engineering problem, just different constraints. They had many more engineers to attack the problem than they had data centers to use for training. They were inevitably going to create a different kind of solution. In hindsight, China took an obvious and clever approach that U.S. companies were mostly blind to because of the path that got them to where they are today before AI.
To put it another way, Nvidia represents one school of thought when it comes to how to build an LLM infrastructure. DeepSeek is an indication that if you aren’t captured by that one school of thought, you have room to innovate. If you are captured by it, innovation is hampered.
What DeepSeek actually shows us is that China, Russia, and many other rival nations are not stopping the development of their own advanced AI systems, and that the U.S. has no choice other than to try to stay ahead of them.
Now, this doesn’t mean we don’t need much more AI infrastructure. In fact, after DeepSeek’s latest release, the company’s servers reportedly crashed because of use, while ChatGPT use went up simultaneously. This means we need more infrastructure, more players in the ecosystem, and more approaches that don’t tie innovation to one path.
It is clear that AI is giving us a critical choice, and we’ve seen this play before in other sectors. We can either build a robust, interoperable AI infrastructure from the ground up and create the conditions for western-allied companies to innovate, or risk repeating past mistakes that left us vulnerable to foreign adversarial technology.
President Donald Trump has already laid the groundwork for this comprehensive approach through his recent executive order, which declares American AI dominance a matter of national policy, aimed at promoting human flourishing, economic competitiveness, and national security.
CLICK HERE TO READ MORE FROM THE WASHINGTON EXAMINER
Rather than accepting the piecemeal and potentially harmful regulations of the previous administration, Trump has ordered a full review of Biden-era AI policies while directing his senior officials to develop a comprehensive action plan by July. This decisive action, including the revision of federal AI governance and procurement guidelines, creates the perfect opportunity to implement an all-of-the-above infrastructure strategy that will secure American leadership in AI.
The stakes couldn’t be higher. AI will be the foundation of economic and military power in the coming decades. By building a robust, interoperable infrastructure now, America can ensure it remains the world’s leading AI superpower while preventing strategic competitors like China from exploiting vulnerabilities in our technological base.
Nathan Leamer is the CEO of Fixed Gear Strategies, a boutique tech policy consulting firm. Previously he was a policy advisor to former Federal Communications Chairman Ajit Pai and also worked as a legislative aide on Capitol Hill.
