Nvidia has stopped shipping licensed Hopper data-center chips, including H200s, to China, despite U.S. approval for some sales. Interim CEO Colette Kress said the company has no current plans for further shipments and its forecast excludes China data-center revenue. Nvidia sold $2.4 billion of licensed Hopper products there in the quarter ended Oct. 25âabout 2% of revenue. Reports say roughly 600,000 H200 chips are awaiting Chinese approval in Hong Kong; China has also restricted imports and favored domestic suppliers.
The halt comes alongside tighter U.S. scrutiny of advanced chips, including rules issued Nov. 16, 2026, that also cover overseas data centers serving Chinese customers remotely. Jefferies estimates these restrictions could reduce Nvidia data-center revenue by $44 billion to $62 billion in 2027. Nvidia reported quarterly revenue of $115.8 billion and forecasts $130 billion in fiscal fourth-quarter revenue, plus or minus 3%, with no China compute sales included. Its board is forming a committee on national security and export controls. President Donald Trump said China might get Nvidia AI chips in two or three years; his administration approved H200 sales but not Blackwell, while the proposed AI Overwatch Act would block Blackwell sales to China for at least 18 months.
Other AI developments span software, infrastructure and investment. OpenAI now requires teams seeking access to gated cybersecurity models to verify their identity, specify the systems they will test and receive approval. Cognition says its Devin agent moved 142 objects in eight days during a customer migration, and reports one NetSuite project took two and a half weeks instead of a planned four. Devin for Databricks Migrations is available on Databricks Marketplace, with purchases counting toward Databricks Universal Commit; Cognition has not disclosed error rates, rework or comparative costs.
Starburst introduced Context Studio, which assembles business definitions from catalogs, BI tools, metric stores and documents to help AI agents answer questions, while retaining access controls and recording source data and SQL. It has not yet published accuracy or productivity results. J.P. Morgan Payments and Hyundai are piloting AI agents for business-to-business payments on a global, multi-currency platform. Separately, Ark Investâs Frank Downing discussed the firmâs early Anthropic investment, while Lightspeedâs Ravi Mhatre discussed Anthropicâs IPO and the importance of public listings for frontier AI firms.
Company results and funding show continued spending and experimentation. Tencentâs third-quarter revenue rose 8% to RMB207.3 billion, but non-IFRS operating profit fell 6% to RMB68.1 billion as AI investment grew; capital expenditure payments jumped 225% to RMB65.1 billion. QIA chief Mohammed Al-Sowaidi said a valuation dip would not alter the fundâs long-term plans. QIA manages $580 billion and has committed $500 billion to U.S. investment over the next decade, despite war-related damage to Qatarâs energy infrastructure and reduced LNG exports.
Vectorial AI raised $3 million in seed funding for its simulated consumer-research products. It says SAPIENS has modelled more than 2.5 million users, run over 30,000 simulations and achieved 87% accuracy in predicting behaviour, at about one-twentieth the cost of traditional studies. Hatz AI raised $24 million in Series A bonds for its platform serving managed service providers and small and midsize businesses. KLIC reported FY26 revenue of $1.183 billion, up 80.9%, with AI and data-center demand driving growth; it also cited customer-concentration and trade-control risks.
Key Takeaways
- Nvidia ended licensed Hopper and H200 shipments to China after $2.4 billion in sales in the quarter ended Oct. 25; its guidance includes no future China data-center compute revenue.
- Nvidia forecasts $130 billion in fiscal Q4 revenue, plus or minus 3%; new U.S. rules cover advanced chips at overseas data centers serving Chinese customers, and Jefferies estimates a possible $44 billionâ$62 billion 2027 data-center revenue hit.
- Nvidia is creating a board committee for national security and export controls. Trump said China might receive Nvidia chips in two or three years; Blackwell sales remain unapproved, and the proposed AI Overwatch Act would block them for at least 18 months.
- OpenAI requires identity checks, defined testing targets and approval before teams can access its gated cybersecurity models.
- Cognition says Devin completed a migration task involving 142 objects in eight days; its Databricks migration product is on Databricks Marketplace, but error rates, rework and comparative costs remain undisclosed.
- Starburstâs Context Studio brings business definitions together for AI agents while retaining access controls and recording sources and SQL; measured results are not yet available.
- J.P. Morgan Payments and Hyundai are piloting AI agents for business-to-business payments on a global, multi-currency platform.
- Investors discussed Anthropicâs early backing and IPO: Ark Investâs Frank Downing outlined its investment, and Lightspeedâs Ravi Mhatre argued public listings matter for frontier AI firms.
- Tencentâs revenue grew 8% to RMB207.3 billion in Q3, while non-IFRS operating profit fell 6% and capital expenditure payments rose 225% as it invested in AI.
- QIA says its long-term plans remain unchanged, including a $500 billion U.S. investment commitment over the next decade; Vectorial AI raised $3 million, Hatz AI raised $24 million in Series A bonds, and KLICâs FY26 revenue rose 80.9% to $1.183 billion.
Nvidia halts licensed AI chip shipments to China
Nvidia has stopped shipping licensed Hopper data center products to China, interim CEO Colette Kress said during an earnings call. The company sold $2.4 billion worth of these products to China last quarter but has no plans for more shipments. The United States had allowed sales of H200 chips to approved Chinese customers under strict license rules, while Beijing has also restricted imports to reduce reliance on U.S. technology. About 600,000 H200 chips were reportedly being held in Hong Kong while awaiting Chinese approval.
Nvidia ends H200 sales as export rules reach overseas data centers
Nvidia interim CEO Colette Kress said the company will no longer sell H200 chips in China, where the government had already largely refused deliveries. Nvidia expects $130 billion in fourth-quarter fiscal 2027 revenue, plus or minus 3%, and its forecast assumes no Data Center compute revenue from China. Rules issued November 16, 2026, also restrict advanced U.S. chips at foreign data centers that serve Chinese customers remotely. Jefferies estimated the rules could cut Nvidiaâs data center revenue by $44 billion to $62 billion in 2027. The company is creating a board committee that will meet at least quarterly and work with legal, national-security and trade-compliance experts.
Nvidia says it has stopped shipping Hopper AI chips to China
Nvidia said Tuesday that it has stopped shipping artificial intelligence chips from its Hopper family to China. The article also mentions allegations about former CEO Jensen Huang, including alleged affairs and a suspected Chinese spy. It provides no further details about the chip shipments or the companyâs reasons for halting them.
Nvidia halts U.S.-approved AI chip shipments to China
Interim CEO Colette Kress said Nvidia has no current plans to ship more licensed Hopper data center products to China. The company sold $2.4 billion worth of these products in the quarter ended October 25, and Kress said future guidance does not include them. The U.S. had allowed H200 sales to approved Chinese customers, including Alibaba, Tencent and ByteDance, under security conditions, while China later restricted imports. Reuters reported that about 600,000 H200 chips were being held in Hong Kong awaiting Chinese approval. The chips are made for Nvidia in Taiwan by TSMC, and eligible exports were subject to a 25% tariff before shipment to China.
Nvidia stops China chip shipments as export scrutiny grows
Nvidia has stopped shipping previously approved Hopper data center chips to China as U.S. scrutiny of chip exports increases. The company is forming a board committee focused on national security and export controls. Nvidia expects fourth-quarter revenue of $130 billion, plus or minus 3%, above the average analyst estimate of $125.25 billion, and its forecast assumes no data center chip sales to China. Board chair Harvey Jones said national security changes are reshaping the market and that the next CEO will need to manage those challenges carefully.
Nvidia stops licensed Hopper chip shipments to China
Nvidia has stopped shipping licensed Hopper data center processors to China, interim CEO Colette Kress said. The company had U.S. approval to sell some products there and reported $2.4 billion in licensed Hopper sales to China in the third quarter. Nvidiaâs revenue outlook does not include future shipments of these chips to China, and Kress did not explain the decision. U.S. regulators have increased checks on advanced chip exports over concerns about remote access to restricted AI technology. Nvidia also formed a board committee on national security and export restrictions.
Nvidia ends licensed Hopper chip sales to China after $2.4 billion quarter
Nvidia has stopped shipping licensed Hopper data center chips to China, interim CEO Colette Kress said on Nov. 17. The company sold $2.4 billion of these products to China in the third quarter, but its next-quarter forecast excludes data center revenue from China. Nvidia reported quarterly revenue of $115.8 billion and data center revenue of $107.6 billion, while forecasting $130 billion in fourth-quarter revenue, plus or minus 3 percent. New U.S. rules tighten oversight of chip exports to overseas data centers that could give Chinese companies remote access to advanced chips. Nvidia said its board would create a committee focused on national security and export-control compliance.
Nvidia stops shipping Hopper AI chips to China
Nvidia interim CEO Colette Kress said the company has stopped shipping licensed Hopper AI chips to China, despite earlier U.S. approval for sales. The Hopper line includes the H200 and is designed to speed up AI model training. Licensed Hopper sales to China totaled $2.4 billion in the third quarter, about 2 percent of Nvidiaâs revenue, but the companyâs guidance includes no such products. Kress gave no reason for the halt, which comes amid closer scrutiny of Nvidiaâs China business. Nvidia also said its board had formed a panel to oversee that business and reported quarterly revenue of $115.8 billion and profits of $65.3 billion.
Nvidia halts licensed Hopper chip shipments to China
Nvidia says it has stopped shipping licensed Hopper AI data center chips to China, even though the United States had approved sales. The company reported $2.4 billion in sales of these chips to China in the latest quarter. Nvidia did not give a reason for stopping shipments, and its guidance does not include licensed Hopper products for China.
Trump says China could get Nvidia AI chips in two or three years
President Donald Trump said the United States might allow Nvidia to sell AI chips to China in two or three years. He said Blackwell chips could become obsolete in a year and described the H200 chips approved for sale to China as âsemi-obsolete.â The Trump administration approved H200 sales but did not approve Nvidiaâs request to sell Blackwell chips to China. Senators Jim Banks and Elizabeth Warren are advancing the AI Overwatch Act, which would block Blackwell sales to China for at least 18 months. Nvidia reported $2.4 billion in licensed Hopper hardware sales to China in its fiscal third quarter, which ended Oct. 25, before halting those shipments.
Nvidia ends H200 chip sales to China
Nvidia interim CEO Colette Kress said the company will no longer sell H200 chips in China, ending a sales arrangement that allowed shipments despite US restrictions. The company also created a board committee to oversee national security policy, export rules and global trade risks. New US rules restrict advanced chips from reaching foreign data centers that serve Chinese customers remotely. Jefferies estimates the rules could cut Nvidiaâs data center revenue next year by $44 billion to $62 billion, though Kress said strong demand for chips including Vera Rubin would protect overall revenue. China had already largely turned away from H200 shipments and favored its domestic chip industry, while the rules may also close a route for Chinese AI labs to access Nvidia Grace Blackwell chips abroad.
OpenAI puts cyber models behind an approval process
OpenAI is requiring teams to pass a review before they can use its gated cybersecurity models. The article says teams must prove their identity, define which systems they will test and obtain approval. Teams considering Blue should plan for that review, since provisioning access alone will not determine the timeline.
Devin says it moved 142 objects in eight days
Cognition says Devin can convert legacy code, test it in a customer workspace, fix failures and compare source data before a migration. The company says Devin runs hundreds of tasks in parallel, with checks for schemas, row counts, totals and row by row differences before cutover. Zeb CEO Mal Vivek said the tool shortened a planned four week NetSuite migration to two and a half weeks and created 20 tables for an investment advisory firm. Devin for Databricks Migrations is available on Databricks Marketplace, and purchases count toward Databricks Universal Commit. Cognition did not share error rates, post-migration rework or costs compared with human-led migrations.
Ark Investâs Frank Downing discusses its early Anthropic investment
Frank Downing, Ark Investâs director of research for AI and cloud, discusses the firmâs early investment in Anthropic. He also shares Ark Investâs longer-term view of the company and the wider AI and technology sector.
Ravi Mhatre says public listings matter for frontier AI firms
Ravi Mhatre, an early Anthropic investor and Lightspeed partner, joins CNBCâs Halftime Report to discuss Anthropicâs record IPO. He also talks about frontier artificial intelligence models and why it is extremely important for frontier AI companies to be public.
Starburst gives AI agents shared business definitions with Context Studio
Starburst Data introduced Context Studio to help AI agents give more accurate answers using business definitions from data catalogs, business intelligence tools, metric stores and documents. The tool brings those definitions together without moving or rebuilding the data, and approval workflows help resolve conflicts. It also records the data sources and SQL queries behind answers, while keeping existing access controls. Starburst has not yet shared measured results on accuracy or productivity and is collecting benchmark data from private preview customers.
JPMorgan and Hyundai test AI agents for business payments
J.P. Morgan Payments and Hyundai Group are testing AI agents that move money between businesses. Umar Farooq, Global Co-Head at J.P. Morgan Payments, described the pilot as part of a global, multi-currency platform. He said the platformâs transaction volumes are steadily growing.
QIA chief says AI market dip will not shake its long-term plans
Qatar Investment Authority CEO Mohammed Al-Sowaidi said a recent dip in AI company valuations does not concern the fund, which invests with a long-term view. He said a market correction could help stronger companies emerge and give the industry time to win public support. QIA manages $580 billion and has committed to invest $500 billion in the US over the next decade. Al-Sowaidi said that plan has not changed despite the US-Iran war and damage to Qatarâs energy infrastructure, which has reduced LNG exports.
Tencent revenue grows 8% as AI spending weighs on operating profit
Tencentâs third-quarter revenue rose 8% year over year to RMB207.3 billion, while non-IFRS operating profit fell 6% to RMB68.1 billion as the company spent more on AI models and apps. Capital expenditure payments climbed 225% to RMB65.1 billion, and free cash flow was negative RMB69.1 billion. Marketing services revenue increased 19% to RMB43.0 billion, which Tencent linked to AI-driven ad targeting and advertising in Weixin Mini Dramas and Mini Shops. Combined Weixin and WeChat monthly active users reached 1.449 billion, up 2%.
Vectorial AI raises $3 million to expand simulated consumer research
San Francisco startup Vectorial AI raised $3 million in seed funding in October to improve its SAPIENS behavioural models and expand its products. Founded in 2024, it has a 12-member team across the US and India and launched SAPIENS in January 2026. The system uses behavioural data to simulate target customers, and Vectorial says it has modelled more than 2.5 million users and run over 30,000 simulations. The company says SAPIENS can deliver feedback within hours at about one-twentieth the cost of a traditional study, and reports 87% accuracy in predicting behaviour.
AI demand drives KLIC revenue up 80.9% in FY26
KLIC reported FY26 revenue of $1.183 billion, up 80.9%, led by demand related to AI and data centers. Net income reached $182.6 million, and bookings and backlog rose sharply. The company still faces risks from high customer concentration, trade controls, and changing demand.
Hatz AI raises $24 million in Series A funding
Hatz AI raised $24 million in Series A bonds to support its growth and expand its products. The company provides an AI platform for managed service providers (MSPs) and small and medium-sized businesses (SMBs). Its platform aims to automate routine tasks, improve customer service, and increase revenue. Hatz AI says the platform is easy to use and works with existing systems, and it has gained customers and partners.
Sources
- Nvidia halts AI chip exports to China amid US tech rivalry
- NVIDIA Ends H200 Chip Sales in China Amid Tighter Export Rules
- Nvidia says no longer shipping 'Hopper' AI chips to China
- Nvidia stops shipping US-approved AI chips to China
- Nvidia pulls the plug on China AI chip sales amidst US exports scrutiny
- Nvidia Makes Major China Decision After $2.4 Billion in AI Chip Sales
- Nvidia Ends China Sales of Its Licensed AI Chips After $2.4 Billion Quarter
- Nvidia says it will stop selling some AI data center chips to China
- Nvidia halts China AI chip shipments despite US approval
- Trump says U.S. might allow Nvidia AI chip sales to China in two to three years
- Nvidia stops sales of advanced AI chips in China
- OpenAI puts cyber models behind vetting
- Devin claims 142 objects moved in 8 days
- Anthropic early investor Ark Invest on the AI giant's historic IPO
- Early Anthropic investor Ravi Mhatre: 'Extremely important' that frontier AI companies are public
- Starburst adds context support to give AI agents consistent access to business data definitions
- JPMorgan and Hyundai Group test AI agents for business payments
- QIA CEO isnât worried about âblipâ in AI super cycle
- Tencent's revenue rises 8% and misses estimates as AI spending cuts adjusted operating profit 6%
- Can Simulated Users Replace Weeks of Testing Products? Vectorial AI Thinks So
- KLIC: AI-led demand lifted FY26 revenue 80.9%, while trade and customer concentration risks persist
- Hatz AI Raises $24 Million in Series A Bonds to Solidify Its Position as the AI Platform for MSPs and SMBs
Comments
Please log in to post a comment.