Technology industry It takes place in front of the surprising fresh agreement of President Trump with Nvidia. Earlier this week, Trump said that it would allow the company to continue to sell its H20 systems to China in exchange for a 15 -percentage of revenues.
“H20 is outdated. You know, it’s one of these things, but he still has the market,” said Trump at a press conference on Monday. “So we negotiated some offers.”
Unusual and legally dubious The agreement is a striking reversal of the Trump administration, which banned all H20 sales to China at the beginning of this year. The president reportedly changed his mind about the problem after meeting with the CEO of NVIDIA Jensen Huang, who he argued This enabling Chinese companies to buy H20 is not a risk for US national security.
On the one hand, this is a plain story about the president who seems to be under the influence of powerful executive lobbying in the interest of his company. But below the surface is a much more captivating and complicated saga about how we got here.
NVIDIA introduced H20 last year after the US government banned the company of sales a stronger system, H800, China. This movement was part of an ambitious project organized by Biden administrative officials, who believed that the United States must first prevent the development of advanced artificial intelligence.
Over the past few months I have worked closely with Graham Webster, a researcher from Stanford University, who tried to understand how and why the Biden team decided that the United States must, above all, limit China’s access to advanced semiconductors. Today, Wired publishes Graham’s final description about what really happened behind the scenes, based on interviews from over 10 former US officials and political experts, some of which said on condition of anonymity.
“I did this song because the official legal justification for control, army and human rights, of course, has never been a whole story,” Graham told me. “Apparently AI was in the mix and I wanted to understand why in a certain depth.”
Graham writes that several key officials in the White House and the Biden Trade Department “believed that AI was approaching the inflection point-a few or several could give the country the main military and economic benefits. Some believed that self-improvement or so-called artificial general intelligence could be just above the technical horizon.
The Biden team decided to take action. In the autumn of 2022, they presented a wide export control aimed at preventing China from accessing the most advanced systems required to train powerful AI systems, as well as specialized Beijing equipment needed to modernize their own national chipipyacing industry.
This movement was the beginning of a long -term project, which “would transform relations between the two largest powers in the world and change the course of what can be one of the most consistent technologies of generations,” writes Graham.
What hit me in Graham’s history is how many people involved in the Biden export control policy went to other influential positions in the world of artificial intelligence, computers and national security. Jason Matheny, who managed the policy of the White House in the field of national technology and security, is currently the president and general director of Rand, an outstanding think tank, who often serves government clients. Tarun Chhabra, who worked on the National Security Council, is now undergoing a national security policy in Anthropic.