By james art 
The first few days of the Trump administration have made major changes to a number of areas – including the business of artificial intelligence.
First, there was a announcement of StargateArtificial intelligence infrastructure project of $ 500 billion. This is a private endeavor – although President Donald Trump revealed him in the White House and he throws his support behind him.
Then there were Chinese bootable Deepseekwhose chatbot shocks investors when it became the most stressed free application at the US Application Store.
And now Openai – the company behind Chatgpt and much of Stargate – has announced that it has partnered with US national laboratories to provide the government with the model for things such as cybersecurity, energy infrastructure and nuclear security.
Mary Louise Kelly of NPR talks with the chief Global Affairs employee of Openai Chris Lehan about the announcement.
This interview is edited for length and clarity.
Interview accents
Mary Louise Kelly: Start with this new partnership with National Labs. In a few sentences why is this a big deal? And what is the time frame for bearing fruit?
Chris Lehane: In short, Openai brings its leading innovation to EDGE and AI technology in the National Labo of the United States. This will be crucial as these laboratories continue to be ahead in research and developments that will support the imperatives of national security. But at the end of the day, what it really brings is our leading EDGE technology, the innovation we build and make sure we do it in cooperation and in partnership with the US government.
Kelly: Returning to Stargate, this huge new joint venture of $ 500 billion, which was just announced. Openai is a key player at Stargate. What is Stargate and why do we need it?
Lehane: Stargate is an infrastructure. At the end of the day, the US and the People’s Republic of China are in a competition, a competition to see who ultimately leads to AI. Currently, the United States has a lead. This is not a huge lead, but we have a lead. Bets could not be more bigge: will we build the world of democratic, free AI, or will it be an authoritarian, autocratic AI? And what will decide this is something called Compute. And Compute is if you add the following pieces: chips, data, talent and energy. And if you put this full stack together, it is an infrastructure of AI from the 21st century. And so what we announced last week with President Trump, with our partners at Oracle and Softbank, is an investment of $ 500 billion in US infrastructure to generate this calculation, which will help ensure the US maintain its leading role In AI.
Kelly: It sounds sublime. Will Stargate create something useful for the average American?
Lehane: First of all, there will be hundreds of thousands of jobs created only by the construction of it. But it also means that the technology generated by AI will be available in a much worse way for many Americans. Consider education – bringing AI to schools to give teachers to help teaching children even better. Consider healthcare, the ability to make sure that people have access to healthcare and the ability to diagnose diseases, challenges, illnesses at a much worse stage.
Kelly: Should it be so expensive? $ 500 billion is a lot of money and Deepseek, the Chinese AI launched, sparked on whether foreign rivals do things much more expensive than US companies.
Lehane: The big absorption for us is that it is absolutely critical that the US has infrastructure so that they can maintain its leading role. So when you talk about $ 500 billion, it’s not taxpayer money in the US. This is money from the private sector, as there is an understanding that since AI continues to move forward, there will be more and more demand for AI, which then requires more and more calculations.
Kelly: Deepseek Detonned Chatgpt, your AI app, this week as the most stressed free application at the US Application Store. Your company says it is investigating whether Deepseek has been used inappropriate Openai data. Where is this?
Lehane: We are in the process of reviewing what may or may not happen. There is something that happens in the world of AI called distillation. And distillation is a complicated idea. But can people effectively send many things to your models and be able to get information from it and then use this information to repeat something else?
Kelly: I wanted to ask what is the difference between a competitive company like Deepseek, which possibly access to your data, and you teach your own products such as Chatgpt for the work of other people without their permission?
Lehane: Maybe a way to think about it is: I go to the library, I take a book, I can read this book and return the book. I have certain information. If you go to the library, pick up the book, never return the book, place the book under your name – these are two different ways to potentially think about it. But I think the big point – the northern star I would just emphasize – is that there is real competition between the US and China. And the consequences and bets of this competition are huge.
Copyright 2025 NPR