How we win.
By Ken Glueck, Executive Vice President, Oracle—Jun 3, 2026
If Congress wants the US to take 2nd place in the global AI race then passing the Remote Access Security Act is an excellent place to start.
The decisions Congress makes during the next few months can set a course for the US to dominate global AI or squander our lead and put the US on a path to second place. These decisions are generational and have profound implications for economic leadership, military dominance, and geopolitics. Controlling the ecosystem upon which AI will be deployed will be as historically and strategically important as establishing the US dollar as the global reserve currency after World War II.
Like the Internet itself, AI is not a single product. Rather, AI is a distributed set of services based on a technology ecosystem, standards, and networks of data centers and cloud infrastructure around the world. The US never owned the Internet but drove its direction through adoption of underlying standards and technologies.
AI is also not a niche technology market where there is significant protectible intellectual property. Most of the underlying technology for AI was commoditized years ago – including the servers, networking, and memory required for AI. Even in areas where the US leads – for example in GPUs, or AI models – that lead can be measured in months, not years. US competitors are perhaps a product cycle or two behind at best. That means AI cannot be controlled within the traditional framework of export controls which presume a scarcity of competing alternatives.
Countries and enterprises are moving quickly to adopt AI, and “good enough” AI provides products and services that are more than sufficient to gain a foothold in the global market. What AI technology vendors understand is we compete in a game of scale, not scarcity, where early and unfettered leadership will turn into enduring technological, economic, and geopolitical advantage. We need to build the equivalent of the Interstate Highway System but in a few years, not decades. There will be no second bite at this apple. The race to reach Promontory Summit and drive the (presciently named) Golden Spike will be as important in AI as it was for commerce more than a century ago.
The problem with AI export controls is that Huawei and DeepSeek already exist, among many other Chinese technologies and models. For that reason, the technologies and infrastructure that scale first win the race. David Sacks, President Trump’s former AI Czar, argued that America’s success should be measured by market share. As he put it, if in five years American technology represents “80% of global compute,” then “that’s winningi.” AI leadership will not be secured by keeping American technology locked inside our borders, particularly when Chinese competitors are already in the market. Leadership will be secured by ensuring the world runs on American infrastructure – our chips, data centers, cloud services, developer tools, applications, and AI models.
We simply cannot afford to slow down at this pivotal moment – a commercial reality lost in most current policy discussions. The global AI market is developing now. Countries and companies around the world are racing to deploy AI infrastructure across healthcare, banking, logistics, manufacturing, energy, transportation, education, and government services. They are doing so now. Enterprises are making platform decisions now. Governments are selecting long-term infrastructure partners now.
AI-driven demand for compute is exploding at an unprecedented rate globally. A massive amount of private capital is required to build out AI infrastructure to meet this demand. If US firms are denied the revenue to build the massive data centers required to drive AI, our competitors will build them instead and dictate the technical standards and the ecosystem that AI operates within. If US firms are prevented from meeting demand, our competitors will fill the vacuum. If US firms are locked out from large parts of the global market due to overly broad US policy “solutions” then, put simply, China will win this race.
We forget history at our peril. We’ve learned the hard way what failure to scale means for an industry that is critical for national security. Huawei’s dominance in global telecommunications, particularly in fifth generation wireless (5G), is a warning about the consequence of US policy failure for AI. Huawei gained global market share through aggressive pricing and state-backed financing, allowing it to establish deep infrastructure footholds across emerging and developed marketsii. Huawei benefited from PRC-sponsored adoption at scale. What followed was both revenue and ecosystem lock-in that helped Huawei steadily improve its technology while making it economically impossible for customers to later switch to Western alternatives. Today’s global telecommunications infrastructure is largely supplied and controlled by China. As policymakers around the world have discovered, “ripping and replacing” the telco installed base once it is established by our competitors is a hope not a strategy.
For these reasons, America’s AI policy should ensure US companies achieve scale first. China’s AI technology does not need to be better than the most cutting-edge US AI technology. Current attempts by Congress to broadly limit the deployment of US AI infrastructure in fact work to advance Chinese interests. To be clear, export control laws and regulations are an essential tool to address national security concerns; however, they work best when they are targeted, predictable, and aligned to support a dominant US AI market share. US policy should continue to restrict the most advanced capabilities from adversary nations’ militaries, intelligence services, sanctioned entities, and diversion networks. But they need to stop there.
The Senate is considering S. 3519, the Remote Access Security Act (RASA). The proposed legislation is so overbroad that it virtually ensures US failure in this highly consequential, generational competition. This legislation will inadvertently hand over a significant portion of the global market to Huawei and other Chinese infrastructure providers. The legislation ensures that US AI infrastructure is locked out of somewhere between 30 to 45 percent of the global market before we even get started, making US dominance mathematically impossible.
RASA also creates inconsistent outcomes that make little sense. It prohibits activity with customers owned by a China headquartered parent company in Singapore or the UAE, while permitting the identical activity with the identical customers in Shenzhen and Ashburn, Virginia. It reaches into allies of the US, like the UK and Germany, prohibiting many hyperscale computing activities for customers far from China. It abandons any existing end-user or end-use controls, and instead draws hard lines around large swaths of the global economy.
The proposed legislation doesn’t mention GPUs or compute capacity. Nor does it mention Large Language Models or model weights. Rather it prohibits all cloud services no matter how benign the underlying technology; prohibits all Chinese customers no matter how benign the business; and then prohibits any other customer that merely employs a Chinese national. Imagine a statute that told Apple or Tesla or Boeing or Lily or Exxon that they are prohibited from doing business with China? So rather than fight for Chinese market access as we have done for decades, we are prohibiting en masse market access for US cloud hyperscalers.
If policymakers want to ignore the lessons of the Internet and 5G, then passing RASA remains an excellent place to start. If the US wants to ensure that US infrastructure never reaches the scale needed to dominate global AI markets, then RASA is exactly the right tool for the job. In short, RASA restricts the reach of the same cloud services that advance the American AI stack and is counter to President Trump’s direction via the AI Action Planiii. It guarantees we lead from behind.
The United States has the best companies, the best chips, the best models, the deepest capital markets, and the strongest innovation ecosystem. But we cannot win if our own rules make American companies slower, less reliable, or harder to buy from than our Chinese competitors. Securing global market share in the AI era is not merely a commercial objective. It is the foundation of American technological leadership, economic strength, and national security. If American companies become the world's preferred AI providers, then the world's infrastructure, standards, developer ecosystems, and governance norms will be built on American technology – entrenching American leadership through scale.
Or, we can be on the outside looking in.
i https://fedscoop.com/white-house-ai-czar-david-sacks-regulations-china/
ii https://www.atlanticcouncil.org/in-depth-research-reports/report/forging-the-5g-future-strategic-imperatives-for-the-us-and-its-allies/
iii https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf