Brian McCullough (0:04)

Welcome to the TechMe Write Home for Tuesday, July 15, 2025. I'm Brian McCullough. Today, has the U.S. government suddenly allowed Nvidia back in business in China, a segment wherein I make the case that gigawatts are the new metric that matters most to the tech industry. Over and above everything else, Windsurf finds a permanent home and is China already producing the smart glasses Zuck wants to see next? Here's what you missed today in the world of tech. Nvidia says it plans to resume H20AI chip sales to China after the US assured its shipments of those chips will be approved, thereby reversing an earlier Trump administration stance from back in April, quoting Bloomberg. US Government officials told Nvidia they would greenlight export licenses for its H20 artificial intelligence accelerator, the company said in a blog post on Monday, a move that may add billions to Nvidia's revenue this year, restoring its ability to fulfill orders it had written off as lost due to government restrictions. Nvidia designed the less advanced H20 chip to comply with earlier China trade curbs from Washington, which Trump's team tightened in April to block H20 sales to the Asian country without a US permit. AMD received similar assurances from the US Commerce Department and plans to restart shipments of its Mi 308 chips to China once licenses for sales are approved, the company said in a statement Tuesday. Shares of AMD jumped as much as 8.5% after markets opened in New York, while Nvidia rose as much as 5%. Nvidia Chief Executive Officer Jensen Huang, who met with President Donald Trump last week and is currently in Beijing attending a government sponsored conference, appeared on Chinese state broadcaster CCTV shortly after Nvidia announced the decision, saying the company had secured approval to begin shipping. The Commerce Department, which oversees U.S. export controls on chips and the tools used to make them, did not immediately respond to a request for comment on whether the agency has already issued any H20 licenses. The US move comes after weeks of thawing relations between Washington and Beijing, guided by an opaque trade truce that's designed to see both sides approve exports of crucial technologies. After meeting his Chinese counterpart last week, US Secretary of State Marco Rubio said there's a strong desire on both sides for a meeting between President Trump and President Xi Jinping later this year. Washington in recent weeks has lifted a spate of export controls, including on chip design software imposed ahead of last month's talks in London. That's in return for China allowing more sales of rare earth minerals needed to make a range of high tech products, something US Negotiators thought they'd achieved the month prior during talks in Geneva. Throughout and after those negotiations, Trump's team insisted that controls on Nvidia's H20 chips were not up for discussion. Treasury Secretary Scott Besant acknowledged Tuesday that the restrictions on Nvidia's H20 chips were part of the London talks, despite his own earlier assertions that there was no such quid pro quo tying semiconductors and rare earths, end quote. Again, I want to underline this isn't just Nvidia that is benefiting from this change. Reportedly, AMD said as mentioned, it will restart its shipments of My308AI chips to China after receiving similar assurances from the US Department of Commerce. But who knows what these things the US Commerce Department this morning apparently opened national security probes into imports of unmanned aircraft systems. So read drones and polysilicon supply chains used in chip manufacturing. So would China be pissed if their drone industry is hit and would then come back on something else which would then reverberate back to chips? Who knows, obviously Related Apple says it assigned a $500 million deal to buy rare earth minerals from US producer MP, which apparently secured some sort of Pentagon backing last week. This is obviously an attempt by Apple cut its reliance on China for some rare earth minerals, quoting Bloomberg the two companies will build a factory in Texas with neodymium magnet manufacturing lines tailored for Apple products, the iPhone maker said Tuesday. In a statement, Apple said the spending on rare earth minerals is part of its earlier pledge to invest more than $500 billion in the US over the next four years. The world's dependence on China for rare earth permanent magnets that are essential for consumer tech cars, wind turbines and fighter aircraft have become a flashpoint in the Asian nation's trade war with the US after the Trump administration imposed 145% tariffs on China. Boasting that it had the upper hand, Beijing turned the tables by essentially shutting down exports of the critical component. MP Materials operates the sole US Rare earth mine at Mountain Pass in California. The increased production will support dozens of new manufacturing and R and D jobs, Apple said. The two companies will also work together to establish a rare earth recycling line in Mountain Pass, California, and develop novel magnet materials and innovative processing technologies to enhance magnet performance, according to the statement. That facility will allow MP Materials to take in recycled rare earth feedstock and use it in Apple products. Magnet shipments from the MP materials facility in Fort Worth, Texas are expected to begin in 2027 and ramp up to support hundreds of millions of Apple devices, the Las Vegas based firm said. China curbs have reverberated across global supply chains. Ford and Suzuki idled some automobile production. El Musk said shortages were hurting his robotics business and governments rushed to secure the few suppliers outside of China. End quote Meanwhile, Google has signed two 20 year power purchase agreements worth $3 billion to access hydroelectric power from Brookfield's renewables arm. It will deliver up to 670 megawatts of power when up and running. CoreWeave says it plans to invest up to $6 billion to set up a 100 megawatt data center in Lancaster, Pennsylv, creating 175 jobs and potentially expanding to 300 megawatts in the future. Are you sensing a trend here? Megawatts are the new metrics that can move markets Mark Zuckerberg says Meta is building several multi gigawatt clusters, starting with one called Prometheus, a massive data center that is scheduled to come online in 2026, again moving from megawatts to gigawatts. Spouting Bloomberg, Zuckerberg on Monday described the data centers under development as multi gigawatt clusters, which would rank among the largest in the world. Meta is building its biggest facility in Richland Parish, Louisiana, which the Facebook co founder has touted as being nearly the size of Manhattan. While most data centers today house only hundreds of megawatts of capacity apiece, several AI and major technology companies like OpenAI and Oracle are involved in plans to develop facilities capable of handling several gigawatts, or roughly enough energy, to power nearly 900,000 homes annually. According to Carbon Collective, that scale of power is seen as necessary to achieve superintelligence. In his post, Zuckerberg cited analysis firm Semianalysis for saying Meta is on track to be the first with a supercluster that houses more than a gigawatt of capacity. In his post on Monday, Zuckerberg reaffirmed that the company is going to invest hundreds of billions of dollars into data center capacity to build superintelligence. We have the capital from our business to do this, he wrote. End quote. And related to that, sources are telling the Times that Meta's new superintelligence lab, led by Alexander Wang, has discussed abandoning its top open source model behemoth in favor of developing a closed model. Quote for years, Meta has chosen to open source its AI models, which means it makes the computer code public for other developers to build on closed models keep their underlying code private. Meta executives have long argued it is better for the technology to be built in public so that AI development will move faster and be accessible to more developers. Any move toward a closed AI model would be a philosophical change at Meta as much as a technical one. Meta has won plaudits from developers for open sourcing its AI models, and one of its top AI executives, Yann LeCun, had said, quote, the platform that will win will be the open one. This year, the Chinese AI company Deepseek released an advanced AI chatbot thanks in part to Meta's open source code. Meta had finished feeding in data to improve its behemoth model, a process known as training, but has delayed its release because of poor internal performance, said the people with knowledge of the matter who were not authorized to discuss private conversations. After the company announced the formation of the Superintelligence Lab last month, teams working on the behemoth model, which is known as a frontier model, stopped running new tests on it, one of the people said. The Superintelligence Lab's discussions are preliminary and no decisions have been made on potential changes which would need sign up from Mark Zuckerberg, Meta Chief Executive. Meta could keep its open source AI models while prioritizing a closed model at the same time. If these scenarios happen, they will be a significant shift for the company as it tries to stay competitive in the AI race against rivals like Google, OpenAI and Anthropic. A Meta spokesman declined to comment on the Superintelligence Lab's discussions. In a podcast interview last year, Mark Zuckerberg said, we're obviously very pro open source, but I haven't committed to releasing every single thing we do. On Tuesday, Mr. Wang held a question and answer session with Meta's AI workers, who number about 2,000. In the meeting, he said the work of his small team would be private, but Meta's entire AI division would now be working toward creating Superintelligence, the people with knowledge of the matter said. He did not address whether AI models would be open or closed in August at the end of the company's next vesting period, which is when some workers are able to sell portions of their stock. Some employees expect an exodus of AI talent who were not chosen to join Mr. Wang's superintelligence team, one of the people with knowledge of the matter said, end quote.