NVIDIA sheds $1 trillion market value

Nvidia’s stock slide is no longer just a market story. By July 8, the company had lost roughly $1 trillion in market value in less than two months, leaving the shares at their cheapest valuation since before the AI boom, according to Bloomberg. France added a second pressure point on July 9. The country’s competition authority said its long-running probe into Nvidia is nearing completion, with general rapporteur Umberto Berkani telling reporters, “We are nearing the end of the investigation.” China is also reshaping the company’s near-term market. Reuters, citing The Information, reported on July 8 that Beijing plans to let some of the country’s biggest AI companies buy limited quantities of Nvidia’s H200 chips. (msn.com) ### How far has Nvidia’s valuation fallen? (aol.com) Bloomberg said on July 8 that Nvidia’s decline has taken its valuation back to pre-AI-boom levels after the stock shed about $1 trillion in market capitalization in under two months. The report described the shares as the cheapest they have been since before the surge in investor enthusiasm around generative AI. (msn.com) Reuters, in a separate pickup of the move, said the shrinking valuation did not reflect a collapse in Nvidia’s operating position. The company’s graphics processors still dominate the AI data-center market, even as investors have cut the multiple they are willing to pay. (msn.com) ### What is France investigating? The French competition authority said on July 9 that its probe concerns alleged anti-competitive practices by Nvidia. Berkani did not give a date for a final ruling, but said the investigation was close to completion. Grafa and other reports said the inquiry has been running for an extended period as regulators examine Nvidia’s position in AI chips and related commercial practices. (economictimes.indiatimes.com) Public reporting has framed the case as part of broader European scrutiny of how dominant AI infrastructure suppliers work with cloud and software partners. (aol.com) ### Why is Nvidia working with d-Matrix instead of just competing? d-Matrix emerged this year as one of the startups trying to sell specialized inference hardware into a market Nvidia has long dominated. CNBC reported in June that the company had entered full production and said its chip could outperform GPUs on some inference tasks. (grafa.com) On July 8, The Next Web reported that Nvidia and d-Matrix were being combined in a joint inference system. A related d-Matrix announcement said Parasail would deploy d-Matrix accelerators alongside Nvidia Hopper and Blackwell GPUs. The Information described that approach as Nvidia partnering with potential rivals rather than treating every new chip company as a direct threat. (cnbc.com) That framing has not been stated by Nvidia in the materials surfaced here, but the reported arrangement shows Nvidia hardware being used alongside a challenger’s inference chips rather than replacing them. (thenextweb.com) ### What is China changing on H200 sales? China has told companies including Alibaba, ByteDance and DeepSeek that they may soon receive permission to buy some H200 chips, according to Reuters’ report on The Information. The move would ease earlier restrictions, but only in limited quantities. (thenextweb.com) The Information and follow-on reports said total approvals could stay below 200,000 chips, with companies required to specify how many they want and why. That would keep access controlled even as Beijing allows some purchases of advanced Nvidia hardware. ### Who is trying to reduce dependence on Nvidia? (msn.com) DeepSeek is building its own inference chip, according to reports cited in the broader coverage around China’s AI supply chain. Those reports said the project is aimed at reducing reliance on both Nvidia and Huawei for key compute components. The next concrete milestone is in France. (theinformation.com) The competition authority said on July 9 that its Nvidia investigation is nearing completion, but it has not yet published a decision date or remedy schedule. (aol.com) (analyticsindiamag.com)