Nvidia faces fresh uncertainty after CEO Jensen Huang resigned following revelations about two extramarital affairs. The departure rattled investors already weighing concerns about AI spending and higher borrowing costs. Nvidia shares have struggled this year, though they remained about 1,100% above their end-of-2022 level at Wednesday’s close. Huang’s exit also raises questions about the company’s role in the data-center buildout, the AI-chip race and Anthropic’s reported planned $1.4 trillion IPO. Separately, AWS CEO Matt Garman voiced doubts about building AI data centers in space, though the available account gives no further detail on his reasoning.
AI infrastructure demand is also shaping investment and logistics stories across Asia. In Japan, Towa, SAMCO and Shibaura Mechatronics could benefit from demand for AI equipment but face risks from higher wholesale inflation, rising interest rates and funding costs. SAMCO’s market value is about ¥55.8 billion and Shibaura Mechatronics’ about ¥191.7 billion. Companies moving AI hardware include Synnex Technology International, valued at about NT$127.8 billion, J&T Global Express and Tokyo Electron Device, valued at about ¥118.2 billion. J&T handled 2.88 billion parcels in Southeast Asia and 271 million in other markets in the third quarter of 2026.
On the technical front, two NVIDIA DGX Stations run the GLM-5.3 NVFP4 model through one OpenAI-compatible vLLM endpoint, using ConnectX-8 RDMA and tensor parallelism to divide computation across the GPUs. The setup supports multi-token prediction, FP8 KV-cache storage and prefix caching. AIPerf tests on SPEED-Bench Coding recorded a peak 607.8 output tokens per second at concurrency 16; at concurrency 32, throughput was 385.0 tokens per second, with mean request latency of 453.5 seconds.
Businesses are testing AI agents and product strategy in parallel. Sierra says Poppy 1.0 is enabled for all Muse users, and Shopify stores are live for agent traffic. Sierra added 60 design partners, taking its total above 100; they include Affirm, Delta Air Lines, Intuit, Salesforce and Zoom. Examples include Mercury banking tasks, Synchrony financing and rewards tests at more than 500,000 merchant locations, and Turkish Airlines flight and booking support. Poppy uses MCP, OpenAPI, UCP and OAuth controls; full purchases inside chat are still to come. IIM Kozhikode and Emeritus announced a Chief Product & AI-Driven Strategy Officer programme on November 13, 2026, covering topics including Generative AI, Agentic AI, product management and AI governance, alongside a Kellogg Executive Education certificate in AI Strategies for Business Transformation.
OpenAI has disbanded its stand-alone robotics team and moved researchers into broader groups, saying the change is intended to accelerate the work. Researchers join the main research team under Mark Chen, while applied robotics and hardware teams move to the Applied organization led by Greg Brockman. OpenAI says it is developing a humanoid robot and targets a first public demonstration in 2027. In health care, the American Medical Association has proposed that developers, rather than doctors, be accountable for errors caused by AI systems. Supporters say this could prioritize patient safety; critics warn it could weaken physician accountability. The AMA is developing a broader framework for safe AI use.
Other developments underline questions about AI’s value and uses. One commentary questions whether generative-AI companies’ financial returns will justify their high valuations and the intense attention they receive. Researchers behind Mantis 2, meanwhile, report that combining simulated and observed outbreak data improves disease forecasts, particularly where local surveillance records are limited. The model forecasts up to six variables 16 weeks ahead; across 30 tasks it had the strongest overall probabilistic accuracy among tested models and missed about 30% fewer outbreak peaks than the strongest comparison. The researchers are making it public for agencies and researchers.
Key Takeaways
- Jensen Huang resigned as Nvidia CEO following revelations about two extramarital affairs, adding uncertainty for the chipmaker, AI investment and data-center expansion.
- Nvidia shares were still about 1,100% above their end-of-2022 level at Wednesday’s close; investors are also weighing AI spending, borrowing costs and Anthropic’s reported planned $1.4 trillion IPO.
- AWS CEO Matt Garman expressed doubts about AI data centers in space; the available article summary does not specify his reasons.
- Japanese AI-hardware stocks Towa, SAMCO and Shibaura Mechatronics face potential demand upside alongside rising interest-rate and funding-cost risks.
- J&T Global Express reported 2.88 billion Southeast Asian parcels and 271 million parcels in other markets in Q3 2026; Synnex Technology International and Tokyo Electron Device are also tied to AI-hardware logistics.
- Two NVIDIA DGX Stations run GLM-5.3 NVFP4 through one OpenAI-compatible vLLM endpoint; tests peaked at 607.8 output tokens per second at concurrency 16.
- Sierra’s Poppy 1.0 is enabled for all Muse users, and Shopify stores are live for agent traffic; its design-partner group now exceeds 100 and includes Salesforce.
- OpenAI moved robotics researchers into its broader research and Applied organizations, while maintaining plans for a humanoid-robot public demonstration in 2027.
- The AMA proposes holding AI developers accountable for health-care system errors, while debate continues over doctors’ liability and accountability.
- Mantis 2 combines simulated and observed data for outbreak forecasts up to 16 weeks ahead and missed about 30% fewer peaks than the strongest comparison model in tests.
Nvidia chief’s exit shakes investor confidence in the AI trade
Jensen Huang’s resignation from Nvidia after revelations about multiple affairs has shocked the tech world. Investors are concerned about what his departure could mean for the wider AI trade and technology stocks. The article explores why Huang mattered to investors and how his exit may affect confidence in the sector.
AWS chief doubts AI data centers in space
AWS CEO Matt Garman expressed doubts about the idea of building AI data centers in space. The article focuses on his view of this proposal amid a rapid expansion in AI and data center development. It does not provide further details about his reasons in the available text.
Jensen Huang’s Nvidia exit could alter Anthropic’s IPO and chip race
Jensen Huang’s departure from Nvidia could affect Anthropic’s planned $1.4 trillion IPO and the race to supply AI chips. The article examines how changes at Nvidia might reshape Anthropic’s prospects and the broader AI chip market. Anthropic CEO Dario Amodei is pictured with Huang in the article.
Three Japanese AI chip stocks face rising interest rate risks
Higher wholesale inflation in Japan in October and expectations for rising interest rates may raise funding costs and pressure import margins. The article looks at three Japanese semiconductor and AI hardware stocks that could benefit from demand for AI equipment, while facing risks tied to borrowing costs and balance sheets. It discusses Towa, SAMCO, which makes plasma etching, thin film deposition and cleaning equipment, and Shibaura Mechatronics, a semiconductor equipment supplier. SAMCO has a market value of about ÂĄ55.8 billion, while Shibaura Mechatronics is valued at about ÂĄ191.7 billion.
Three logistics stocks tied to the global AI hardware boom
Rising trade in AI hardware is increasing demand for fast, reliable shipping of valuable goods across borders. The article highlights Synnex Technology International, J&T Global Express and Tokyo Electron Device as companies connected to this trend. Synnex distributes technology and semiconductor products and has a market value of about NT$127.8 billion. J&T reported 2.88 billion parcels in Southeast Asia and 271 million in other markets in the third quarter of 2026, while Tokyo Electron Device distributes semiconductors and electronic components and has a market value of about ÂĄ118.2 billion.
Two NVIDIA DGX Stations run GLM-5.3 through one vLLM endpoint
A setup with two NVIDIA DGX Stations uses ConnectX-8 RDMA to run the NVIDIA GLM-5.3 NVFP4 model through one OpenAI-compatible endpoint. vLLM uses tensor parallelism to split the model’s weights and computing across the two GPUs. The service also supports multi-token prediction, FP8 KV-cache storage and prefix caching. AIPerf tests on the SPEED-Bench Coding dataset measured a peak output rate of 607.8 tokens per second at concurrency 16, while the rate dropped to 385.0 tokens per second at concurrency 32, with mean request latency of 453.5 seconds.
Nvidia’s market-leading stock faces new uncertainty
Nvidia’s success in AI chips helped make it the world’s most valuable company and a major driver of stock market gains since 2023. Its shares have struggled this year amid concerns about AI spending, and CEO Jensen Huang resigned Thursday after revelations about two extramarital affairs. Huang’s departure adds uncertainty for Nvidia and the data center expansion, as investors also weigh rising borrowing costs for AI companies and the planned Anthropic IPO. Nvidia’s revenue grew rapidly as technology companies increased AI spending, and its shares were still up about 1,100% from the end of 2022 at Wednesday’s close.
AMA seeks protection for doctors when AI systems make mistakes
The American Medical Association has proposed that doctors should not be held liable for mistakes made by AI systems in health care. Presented to the National Academy of Medicine, the proposal says developers should be accountable when their systems cause errors. The AMA says current rules can unfairly expose doctors to lawsuits over mistakes that are not their fault. Critics warn that a liability shield could weaken doctors’ accountability, while supporters say it could help put patient safety first. The AMA is working on a framework for safe AI use, as debate over the proposal continues.
Sierra brings Poppy 1.0 to Shopify and expands its business tests
Sierra says Poppy 1.0 is now enabled for all users of Muse, which is owned by Meta, and that Shopify stores are already live for agent traffic. The company added 60 design partners, bringing its total to more than 100, including Affirm, Delta Air Lines, Intuit, Salesforce and Zoom. Current examples include Mercury banking tasks, Synchrony financing and rewards tests at more than 500,000 merchant locations, and Turkish Airlines flight and booking support. Poppy 1.0 uses MCP, OpenAPI and UCP, with OAuth controls, and companies can adopt it in stages using Sierra’s implementation kit. Full purchases inside chat are still to come.
IIM Kozhikode and Emeritus launch AI product leadership programme
The Indian Institute of Management Kozhikode and Emeritus announced the Chief Product & AI-Driven Strategy Officer programme on November 13, 2026. It is aimed at senior product professionals, business leaders and entrepreneurs who want to grow organizations through product-led and AI-enabled strategies. Topics include AI-driven marketing, Generative AI and Agentic AI, product lifecycle management, cloud and DevOps, and AI ethics and governance. The curriculum also includes a Kellogg Executive Education online certificate in AI Strategies for Business Transformation. Participants can study how to use tools and methods such as LLMs, NLP, RAG and Computer Vision, and earn credentials from IIM Kozhikode and Kellogg Executive Education, with eligibility for IIM Kozhikode Executive Alumni Status.
Generative AI faces questions about its financial value
The author has warned since August 2023 that generative AI may not deliver enough financial value to match the excitement around it. The article points to high company valuations and widespread media coverage as signs of the intense hype. It questions whether these companies will ultimately be worth far less than their current valuations suggest.
OpenAI shifts robotics researchers into its wider AI teams
OpenAI has disbanded its stand-alone robotics team and reassigned its researchers to other groups. The company said the move aims to speed up robotics work, not end it. Robotics researchers will join the main research team under Chief Research Officer Mark Chen, while applied robotics and hardware teams will join the Applied organization led by President Greg Brockman. OpenAI is betting that general-purpose AI models can control robots, and it says it is developing a humanoid robot with CEO Sam Altman targeting a first public demonstration in 2027.
Mantis 2 improves disease outbreak forecasts using two types of data
Researchers developed Mantis 2, a forecasting model that combines simulated outbreak data with observed data. Their tests found simulated data especially useful when local surveillance history is limited, and it helped reduce very large forecasting errors. Mantis 2 can forecast up to six variables sixteen weeks ahead. Across 30 forecasting tasks, it had the best overall probabilistic accuracy among the machine learning, statistical, and baseline models tested, and it missed about 30% fewer outbreak peaks than the strongest comparison model. The researchers are making the model public for agencies and researchers to forecast disease outcomes without building a separate model for each disease or location.
Sources
- Investors worry about fate of AI trade after Nvidia chief resigns
- AWS CEO throws cold water on AI data centers in space
- How Jensen Huang's fall could reshape Anthropic IPO and AI chip race
- Japanese AI Chip Stocks Worth Watching As Higher Rates Test Balance Sheets
- 3 Stocks Riding The AI Hardware Trade Boom In Global Logistics
- When One GPU Isn’t Enough: Serving GLM-5.3 Across Two NVIDIA DGX Stations with vLLM
- Nvidia Has Been the Market’s Golden Goose. Now It’s in Crisis
- Doctors want liability shield from AI mistakes
- Sierra locks in Poppy 1.0 with Shopify live
- IIM Kozhikode Launches Chief Product & AI-Driven Strategy Officer Programme.
- The great Generative AI unraveling
- Exclusive: OpenAI disbands its robotics team as its general models start running robots
- Universal Infectious Disease Forecasting with Mantis 2
Comments
Please log in to post a comment.