Nvidia’s chief executive Jensen Huang has never been coy about the central role China plays in the company’s future. But his latest comments in Washington this week laid bare a deeper unease inside the world’s most valuable chipmaker.

After meeting U.S. President Donald Trump on Wednesday, Huang confirmed he was still unsure whether Beijing would even allow Chinese firms to buy the company’s proposed H200 artificial intelligence chips, even if Washington greenlights sales.

The uncertainty amounts to an extraordinary moment for a company that built its global AI leadership partly on enormous demand from Chinese cloud giants and research labs. Now, Huang is lobbying the U.S. administration and Congress to loosen export restrictions that have cut off Nvidia’s access to its biggest overseas market.

At the same time, a new crop of Chinese semiconductor players is advancing high-end processors of their own, accelerating a shift that could close the door on Nvidia for years or permanently.

According to Bloomberg, Huang said he had spoken with Trump about export controls during their meeting but did not share specifics. Asked whether China would even accept the H200 if Washington approved the sale, he answered, “We don’t know. We have no clue.” He then made a candid point about market reality, saying, “We can’t degrade chips that we sell to China, they won’t accept that.”

The H200 sits one generation behind Nvidia’s current flagship processors. Washington has blocked the company’s most advanced chips from entering China, citing national-security concerns tied to advanced model training and military use cases. The proposed H200 offering was seen as a possible compromise, but even that pathway is now shaky.

Analysts in Beijing say the discomfort inside Nvidia is real. Ma Jihua, a veteran telecom researcher quoted by the Global Times, said the company’s recent push in Washington shows genuine worry about the pace of China’s ecosystem overhaul. Ma noted that the H200 still leads in performance, memory bandwidth, and its integrated software suite, but the window for Nvidia to maintain that lead in China is closing fast. Several mainland firms are improving process technology and pushing toward top-tier chips that can run large AI models without U.S. hardware.

Huang’s concern is reinforced by his earlier appearance on FOX Business in late November, where he warned that access to Chinese buyers is vital for long-term U.S. AI competitiveness. He said the export restrictions have pushed Nvidia’s China sales to zero for two straight quarters. In his view, the U.S. innovation engine has always depended on enormous market returns, and losing China undermines that cycle.

Zhou Mi, a senior researcher at the Chinese Academy of International Trade and Economic Cooperation, explained this dynamic further in comments to the Global Times. He said the United States risks damaging the same economic structure that supports its technology leadership if it keeps stretching national-security rules across wider categories. According to him, an expanded security doctrine may rattle global supply chains in a way that weakens the base Washington says it wants to protect.

Trump, for his part, spoke only in brief lines about the meeting. When asked whether he had conveyed his stance on control rules directly to Huang, Trump responded, “He knows.” He also described Huang as a “smart man.”

Behind the scenes, Huang has also been trying to influence Congress. One idea under consideration earlier this year would have forced Nvidia to offer advanced chips to U.S. buyers first before seeking approval to sell to “countries of concern.” Nvidia resisted it, warning lawmakers that such requirements could harm the nation’s competitive position in global AI markets. For now, a similar restriction was kept out of major defense legislation.

Huang also delivered another message while meeting lawmakers. According to CNBC, he warned that state-level rulemaking in the U.S. could slow AI progress, since companies would need to navigate multiple compliance frameworks rather than a single federal standard.

Later that day, during an appearance at the Center for Strategic and International Studies, he sought to calm speculation that Nvidia’s large data-center GPUs are being smuggled into restricted countries.

“A GPU for AI data centers, that GPU weighs two tons,” he said. “It has one and a half million parts. It consumes 200,000 watts. It costs $3 million.”

He added that anyone claiming mass smuggling of those systems would need to explain how entire shipments the size of a football field are crossing borders unseen.

The broader landscape around all of this is evolving quickly. Washington’s rules have already reshaped the global AI supply chain, and China’s fast-moving domestic chip race has become one of the biggest variables for 2025 and 2026. Nvidia still owns the high end of the market, but the danger for 2026 is that the company could exit China unintentionally, not through policy but through technological substitution. Beijing’s AI firms are motivated by both necessity and national ambition, and years of restricted access have forced them to accelerate their internal R&D pipelines.

At the same time, Chinese cloud providers are designing around Nvidia’s CUDA software advantages, building their own frameworks or investing in open-source alternatives. Once those systems mature, the lock-in power that elevated Nvidia over rivals such as AMD and Google could lose force. Huang appears aware of this shift and is pressing Washington to avoid creating conditions that push China’s market permanently out of reach.

The political mood in Washington has also hardened in ways that make an easy rollback unlikely. Trump has signaled that national-security concerns remain central to his thinking, and congressional committees across both parties are pushing for tighter oversight of AI hardware exports. For Nvidia, this means the timeline for clarity is uncertain, and each delay gives Beijing more time to scale homegrown options.

The stakes are high because Nvidia’s valuation depends heavily on demand for advanced AI accelerators. The U.S. market is huge, but China historically made up a significant share of data-center GPU revenue. With zero sales expected from there for two consecutive quarters, the pressure on Nvidia will only intensify.

The next year will show whether Washington softens its stance on H200 sales or doubles down. It will also show whether Chinese firms can move from prototypes to mass-market systems that compete directly with Nvidia’s constrained offerings.

However, Nvidia’s latest lobbying burst shows that the company sees this as a make-or-break moment. Huang’s blunt admission that he has “no clue” what China will accept underlines just how unpredictable the ground beneath the AI hardware market has become.

Meta Platforms, the parent company of Facebook and Instagram, is reportedly planning to implement massive budget cuts of up to 30% for its ambitious metaverse initiative, according to a report from Bloomberg News on Thursday.

The news served as immediate relief for investors, sending Meta’s shares surging by nearly 4% in morning trading, as the move signals a major strategic pivot away from a costly long-term bet that has burned more than $70 billion since 2020.

The proposed cuts are part of the company’s annual budget planning for 2026, which included a series of executive meetings last month at CEO Mark Zuckerberg’s Hawaii compound. Cuts of this magnitude, which are expected to predominantly affect the Virtual Reality (VR) group and the Horizon Worlds social platform, would most likely include significant layoffs as early as January, though Meta has not officially confirmed the report.

“Smart move, just late,” said Craig Huber, an analyst at Huber Research Partners. “This seems a major shift to align costs with a revenue outlook that surely is not as prosperous as management thought years ago.”

The metaverse efforts sit within the Reality Labs division, which is responsible for the company’s hardware—including the Quest mixed-reality headsets, smart glasses made with EssilorLuxottica (ESLX.PA)’s Ray-Ban, and upcoming augmented-reality glasses. The division has been a staggering financial drain, accumulating over $70 billion in cumulative losses since early 2021, and reported a loss of $4.4 billion in the most recent quarter alone.

While Meta has struggled to sell its vision of an immersive, interconnected metaverse to the mainstream, the division has achieved an early lead with its Ray-Ban smart glasses, finding a niche where competitors like Alphabet’s Google, Apple, and Snap have previously failed to capitalize. However, this success has not been enough to offset the enormous capital drain from the VR projects.

Shift to AI: The New $72 Billion Priority

The proposed metaverse cuts underscore a critical shift in the company’s focus: the Artificial Intelligence (AI) race. Meta is aggressively attempting to gain ground in Silicon Valley’s AI competition, especially after its highly anticipated Llama 4 model met with a poor reception from the developer and research communities. The model was criticized for being underwhelming on key benchmarks (like coding and complex reasoning), having a restrictive open-source license, and being perceived by some as a rushed release.

To fuel its new AI ambitions, Meta has committed as much as $72 billion in capital spending this year, contributing to the nearly $400 billion in total expected AI expenditure by large tech companies across the industry. The company has reorganized its AI efforts under Superintelligence Labs, with Zuckerberg personally leading an aggressive talent acquisition effort, reportedly floating million-dollar pay packages and directly courting top prospects on WhatsApp.

Centralized Support for Facebook, Instagram

In a separate move, Meta announced the launch of a new, centralized support hub for Facebook and Instagram users, acknowledging that its prior support options haven’t “always met expectations.” The hub aims to centralize account recovery options, offer clearer guidelines, and use new AI systems to help users report issues, recover lost accounts, and get answers via an AI-powered search and assistant.

The company claims its use of AI is improving user safety, noting that account hacks have decreased by over 30% globally across Facebook and Instagram. AI is also used to identify and stop other threats, such as phishing and suspicious logins, and has helped speed up the appeals process for mistakenly disabled accounts.

However, this official claim stands in contrast to the “lived experience of thousands of users” who complain of losing access to their accounts or business pages due to perceived system errors, with many suspecting the lack of human oversight in AI-driven decisions is to blame.

The severity of the issue has led to the creation of an entire Reddit forum dedicated to helping people who are suing Meta over these disabled accounts. Meta believes the new hub, which includes an optional selfie video verification for identity, will address these long-standing customer service problems.