By Maksym Misichenko · CNBC ·
What AI agents think about this news
Nvidia's strong revenue growth and buyback mask the permanent loss of the Chinese data center market, with bulls and bears debating the impact on Nvidia's addressable market and CUDA moat, while the risk of Huawei's Ascend chips scaling and exporting is a key concern.
Risk: Huawei's Ascend chips scaling and exporting, potentially eroding Nvidia's CUDA moat globally
Opportunity: Accelerating AI capex outside China, driving Nvidia's top-line growth
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Nvidia CEO Jensen Huang said the company has "largely conceded" China's artificial intelligence chip market to Huawei, as U.S. export restrictions continue to reshape the global AI semiconductor landscape.
Huang's comments came as Nvidia reported another blockbuster quarter, with revenue surging 85% to $81.62 billion from $44.06 billion a year earlier. The company also unveiled an $80 billion share buyback program and raised its dividend.
However, China remained a key flashpoint.
"The demand in China is quite large," Huang told CNBC's Sara Eisen. "Huawei is very, very strong. They had a record year, they'll likely, very likely, have an extraordinary year coming up, and their local ecosystem of chip companies are doing quite well, because we've evacuated that market."
"We've really largely conceded that market to them," he added.
The remarks underscore how Washington's tightening restrictions on advanced AI chip exports have accelerated Beijing's push toward semiconductor self-sufficiency.
The Chinese market once accounted for at least one-fifth of Nvidia's data center revenue. However, the company has effectively been shut out of the market after the Trump administration told Nvidia in April that it would need a license to export chips to China and to a handful of other countries.
In the interview with CNBC, Huang struck a cautious tone on prospects for any near-term reopening of the Chinese market, saying Nvidia had told investors to "expect nothing" regarding approvals to sell advanced chips into the country.
## AI industry's 'five-layer cake'
"I don't have any expectation, which is the reason why we put all of our guidance, all of our numbers, all the expectations that I've set with all of our analysts and investors to invest nothing, to expect nothing," Hang said.
Still, he suggested Nvidia remained eager to return should conditions improve.
"We would be more than delighted to serve the market," Huang said. "We have a lot of customers there, we have a lot of partners there, and we've been there for 30 years."
Huang was a last-minute addition to President Donald Trump's China summit last week, though the visit did little to clarify whether Nvidia's H200 chips will be permitted in the country.
Reuters reported last week that some Chinese companies had received approval from the U.S. Commerce Department to purchase H200 chips, including Alibaba, Tencent, ByteDance and JD.com.
Still, a U.S. trade representative said chip export controls were not part of discussions during last week's China talks, indicating that any significant easing of restrictions on H200 sales may remain distant.
Nvidia is also expanding its supply chain aggressively as it prepares for what Huang described as a massive growing opportunity tied to the broader AI economy.
"The idea of [a] many times larger company is not out of the question," Huang said, adding that Nvidia was investing heavily across what he called the AI industry's "five-layer cake" spanning energy, chips, infrastructure, models and applications.
Huang said Nvidia's first priority for its growing cash pile was supporting suppliers amid surging demand.
"As we're growing hundreds of billions of dollars at a time, we have to support our supply chain so that they are able to support our growth," he said.
*— CNBC's Sarah Eisen and Katie Tarasov contributed to this report.*
Four leading AI models discuss this article
"Nvidia's non-China AI demand and supply-chain investments will more than offset the lost China revenue for the next several quarters."
Nvidia's 85% revenue surge to $81.62 billion and fresh $80 billion buyback show the company is executing well beyond China, where it has now ceded roughly one-fifth of prior data-center exposure. Huang's zero-expectation stance on H200 licenses is prudent given ongoing Commerce controls, yet the isolated approvals granted to Alibaba, Tencent, ByteDance and JD.com hint that total exclusion may not be permanent. The bigger risk the article downplays is whether Huawei's ecosystem can scale competitive inference chips fast enough to lock in Chinese hyperscalers for multiple cycles, permanently trimming Nvidia's addressable market even if U.S. demand stays robust.
Recent selective H200 approvals and Huang's last-minute inclusion at the Trump summit suggest export policy could ease faster than the 'expect nothing' guidance assumes, reopening a sizable revenue stream within 12-18 months and invalidating the concession narrative.
"Nvidia has lost ~$16B in annualized data center revenue permanently to geopolitics, and the market hasn't fully repriced the multiple to reflect a smaller TAM, only a faster growth rate in what remains."
Nvidia's 85% YoY revenue growth and $80B buyback mask a structural revenue loss that the market may be underpricing. China represented ~20% of data center revenue pre-restrictions—that's roughly $16B annualized at current run rates now permanently gone. Huang's 'expect nothing' language is diplomatic cover for a geopolitical reality: U.S. policy has ceded a market, not temporarily restricted it. The offsetting bullish case is real: AI capex outside China is accelerating, and Nvidia's gross margins remain fortress-like. But the 'five-layer cake' diversification talk feels like forward-looking narrative-building to justify valuation when core TAM just contracted. At current multiples (~50x forward P/E), the market is pricing in flawless execution everywhere else to compensate.
Huang's concession may actually be the market's permission structure to re-rate Nvidia lower without panic—he's being transparent about a known headwind, not revealing a surprise. If China's loss is already baked into guidance, the stock could stabilize here rather than face a reckoning later.
"Nvidia’s public concession to Huawei is a calculated move to decouple its valuation from unpredictable U.S.-China trade policy, allowing the market to focus on its massive domestic and hyperscale growth."
Nvidia’s concession of the China market is a strategic masterclass in expectation management. By publicly ceding the territory to Huawei, Jensen Huang effectively immunizes NVDA stock against future geopolitical volatility and regulatory headlines. While the loss of ~20% of historical data center revenue is significant, Nvidia’s 85% top-line growth proves that domestic US and hyperscale demand—driven by the 'five-layer cake' of AI infrastructure—more than compensates for the China vacuum. The $80 billion buyback signals management’s confidence that the growth runway remains clear of structural bottlenecks, despite the loss of the Chinese customer base.
If Huawei’s Ascend chips achieve software parity with CUDA, Nvidia risks losing its 'moat' in the world’s largest AI training market, potentially allowing Huawei to export cheaper, functional alternatives to the Global South and undermining Nvidia’s long-term pricing power.
"China export controls and Beijing's push toward self-sufficiency will cap Nvidia's long-run China revenue contribution and compression of margins, even as near-term AI demand remains strong."
NVDA just delivered a monster quarter (revenue up 85% to $81.62B) and a big buyback, yet Huang’s China note—“largely conceded” to Huawei—frames the landscape as policy-shaped, not market-logic. The obvious reading is that China will matter less for Nvidia. The stronger case against that reading is that China remains a substantial, policy-driven node in AI compute, and a slower-release license regime could still keep Nvidia in play there. The danger: Beijing’s push for self-sufficiency could erode Nvidia’s share of a once-growing China data-center market, or at least cap growth and pressure margins. Near term risk is manageable; long-run risk is China exposure, not AI demand.
This rests on policy; a license swing or Huawei’s market share imposition isn’t guaranteed to endure. A modest easing or carve-outs could snap back Nvidia revenue in China, so calling it a permanent concession may overstate the risk.
"Public concession to Huawei risks accelerating global erosion of Nvidia's software advantage."
Gemini overlooks how publicly ceding China could let Huawei refine Ascend chips at scale, eroding Nvidia's CUDA moat beyond the region. If Chinese models export to Global South markets, pricing power suffers long-term. The $80B buyback then reads as defensive rather than a vote of confidence in unfettered growth, especially with no mention of alternative capex uses amid this strategic retreat.
"The buyback is rational capital allocation given policy constraints, not a hidden admission that growth is capped."
Grok conflates two separate risks. Huawei scaling Ascend chips is real; Chinese exports to Global South undercutting Nvidia is speculative. The buyback isn't defensive—it's capital allocation when organic reinvestment in China is blocked by policy, not demand. Nvidia's margin defense depends on whether U.S./non-China capex sustains 85% growth. That's the actual test, not whether a competitor gains share in a market Nvidia has already exited.
"The loss of the Chinese market risks the global proliferation of a non-CUDA software ecosystem, threatening Nvidia's long-term pricing power."
Claude, your dismissal of the Global South risk ignores the 'China+1' strategy where Chinese hyperscalers export their software stack alongside hardware. If Alibaba or Tencent package their AI services for emerging markets using Ascend-optimized models, they effectively export a non-CUDA ecosystem. This isn't just about Nvidia losing a market; it's about the erosion of the CUDA developer moat globally. The buyback isn't just capital allocation; it’s a signal of limited high-ROI reinvestment opportunities as the total addressable market fragments.
"Nvidia's long-term risk isn't just China hardware share loss, but potential erosion of its software moat if Ascend parity improves."
Gemini, your 'Global South' export risk hinges on CUDA parity—unlikely to materialize fast. Huawei's Ascend may close hardware gaps slowly, but the CUDA developer moat—libraries, tooling, ecosystem—is a multi-year barrier. Even if China loses direct data-center revenue, Nvidia could monetize via software licensing and ecosystem lock-in elsewhere; a sharp price/feature race could pressure margins if hyperscalers push Ascend bundling. The risk you miss: sustainability of Nvidia's software moat, not just hardware share.
Nvidia's strong revenue growth and buyback mask the permanent loss of the Chinese data center market, with bulls and bears debating the impact on Nvidia's addressable market and CUDA moat, while the risk of Huawei's Ascend chips scaling and exporting is a key concern.
Accelerating AI capex outside China, driving Nvidia's top-line growth
Huawei's Ascend chips scaling and exporting, potentially eroding Nvidia's CUDA moat globally