AMD and Marvell both rebound after an AI-stock selloff, but their investment cases differ. AMD reports record third-quarter revenue of $13.4 billion and has major chip commitments from OpenAI, Meta and Anthropic, though investors weigh its high share price, export restrictions and pricing pressure. Its average analyst price target reaches $657, while its 2027 earnings estimate rises to $17.75 per share from $16.28 on Oct. 30. Marvell posts record fiscal second-quarter revenue of $2.74 billion and raises future sales goals; its proposed $115 billion stock purchase of Lumentum could strengthen its position in lasers and optical technology, but the deal and customer concentration carry risks.
Jensen Huang’s reported departure from Nvidia unsettles markets and prompts questions about the company’s longer-term direction. Analysts say AMD could win GPU orders and Broadcom could benefit from custom AI-chip work with Meta, while Alphabet’s tensor processing units may draw more customers. Daol Investment & Securities says the immediate effect on the semiconductor cycle is limited, but Nvidia’s post-Feynman plans beyond 2028 will depend on whether new leaders maintain strategic consistency. Nvidia is due to report earnings Nov. 17. Its shares close at $166.19, down from $173.73 the previous day; at 10.9 times estimated 2027 earnings, it is the second-cheapest among the listed AI companies, behind Micron at 4.3 times. AMD trades at 33.2 times.
Market concerns extend beyond leadership changes. Asian markets steady after the Nvidia news, while the dollar rises alongside Treasury yields. Investors also monitor tighter U.S. chip-export limits, weaker copper demand and a growing possibility of a 50-basis-point Federal Reserve rate hike. Uncertainty over AI projects could raise financing costs for data centers, and traders are reassessing whether AI construction will boost near-term demand for copper and other industrial metals.
AI applications are also moving into logistics and financial services. EAIGLE says its AI camera system can cut truck check-ins at distribution centers from about 10 minutes to under 30 seconds, potentially reducing diesel use and trucking costs. The U.S. average diesel price was $6.13 a gallon for the week of Nov. 9; any savings passed through supply chains could ease some pressures behind grocery prices. Separately, a Finance Magnates Summit London 2026 session on Nov. 24 will examine how trading firms can move from AI copilots to autonomous systems, including data quality, permissions, accountability and control of agent decisions.
Researchers have proposed a hybrid AI model that uses a quick answer when deeper reasoning is not worth its cost, and routes harder tasks to step-by-step reasoning. Its lightweight router is trained with Calibrated Decision Policy Optimisation for agent tasks that lack labels for each decision. On JevBench, HDM-4B scores 0.857 versus Jev’s 0.851 and has a median time of 0.179 seconds versus 0.652; on MiniWoB, HDM-27B cuts token costs by 80.1%.
Two other debates concern AI infrastructure and governance. Microsoft has offered buyouts while investing heavily in AI infrastructure, having projected $10 billion in AI investment earlier this year. No evidence links its workforce cuts or AI spending directly to Thursday’s outage, though the disruption raises questions about whether reorganizing and reducing staff could leave fewer resources to maintain existing systems. U.S.-China AI safety talks face different definitions of serious incidents: Washington emphasizes loss of control over AI, while Beijing prioritizes stability and content rules. The governments have established a dialogue mechanism ahead of the APEC summit.
Anthropic CEO Dario Amodei faces growing opposition from technology leaders over his warnings about powerful AI models. Anthropic and OpenAI urged the White House to restrict development and sharing of some models, while Microsoft CEO Satya Nadella and then-Nvidia CEO Jensen Huang supported keeping them open. A letter backing open models drew two dozen company signatories and later 200 more, including OpenAI; Amodei declined to sign. The dispute unfolds as Anthropic reports strong revenue growth, takes a leading role in AI development and moves toward a possible major public offering, amid criticism of its business practices and actions against free alternatives.
Key Takeaways
- AMD reports record third-quarter revenue of $13.4 billion and has major chip commitments from OpenAI, Meta and Anthropic; its 2027 earnings estimate rises to $17.75 per share.
- Marvell records $2.74 billion in fiscal second-quarter revenue and proposes a $115 billion stock purchase of Lumentum, with deal and customer-concentration risks.
- Nvidia’s reported CEO departure raises questions about its long-term GPU roadmap; Daol says near-term semiconductor-cycle effects are limited, while plans beyond Feynman depend on new leadership.
- Analysts see potential for AMD to win GPU orders, Broadcom to sell custom AI chips to Meta, and Alphabet’s TPUs to attract customers; Nvidia reports earnings Nov. 17.
- Nvidia trades at 10.9 times estimated 2027 earnings, versus AMD at 33.2 times; Nvidia closes at $166.19 after falling from $173.73.
- EAIGLE says its AI cameras can shorten distribution-center truck check-ins from about 10 minutes to under 30 seconds; U.S. diesel averaged $6.13 a gallon for the week of Nov. 9.
- A proposed hybrid reasoning model scores 0.857 on JevBench versus 0.851 for Jev, runs faster in the reported median-time comparison, and cuts HDM-27B token costs by 80.1% on MiniWoB.
- Microsoft’s workforce buyouts and AI investment coincide with an outage, but the article reports no evidence that either caused it; the company earlier projected $10 billion in AI investments.
- U.S.-China AI safety talks face differing priorities, while a new dialogue mechanism is in place ahead of the APEC summit.
- Anthropic CEO Dario Amodei’s push for limits on some powerful AI models draws opposition from leaders backing open models; a supporting letter grows from two dozen companies to 200 more signatories, including OpenAI.
AMD and Marvell offer different AI investment opportunities
AMD and Marvell shares both recovered after a recent AI stock selloff, but AMD rose more. AMD reported record third-quarter revenue of $13.4 billion, while Marvell reported record fiscal second-quarter revenue of $2.74 billion and raised its future sales goals. Investor Robert Izquierdo says AMD’s growth and new customer deals come with a high share price, while Marvell may gain from its planned $115 billion stock purchase of Lumentum, a maker of lasers and optical technology. Both companies face risks, including export limits, reliance on a small number of partners or customers, and uncertainty around the Lumentum deal.
Nvidia leadership turmoil shifts attention to AMD and Broadcom
Investors are reacting to Jensen Huang’s departure from Nvidia, with AMD and Broadcom shares rising while CoreWeave and Dell Technologies shares fell. Analysts see a chance for AMD to win GPU orders and for Broadcom to gain sales of custom AI chips, though they warn that markets can overreact and that Nvidia’s earnings report on Nov. 17 may offer more information. AMD has large chip commitments from OpenAI, Meta Platforms and Anthropic, but possible China export limits and pricing pressure remain risks. Broadcom could benefit from its work with Meta on custom MTIA chips, while Alphabet’s tensor processing units may also attract customers. AMD’s average price target is $657, and its 2027 earnings estimate has risen to $17.75 a share from $16.28 on Oct. 30.
AI cameras aim to cut truck delays and shipping costs
EAIGLE says its AI-powered camera technology can cut truck check-ins at distribution centers from about 10 minutes to under 30 seconds. Faster check-ins could reduce diesel use and costs for trucking companies. The article notes that the U.S. national average diesel price was $6.13 per gallon for the week of Nov. 9. If savings spread through supply chains, they could help ease some of the cost pressures behind grocery prices.
Daol Securities sees uncertainty in Nvidia’s future plans after Huang’s exit
Daol Investment & Securities said Jensen Huang’s abrupt resignation as Nvidia CEO adds uncertainty to the company’s medium and long-term GPU roadmap and its investment strategy for the AI ecosystem. Analyst Nam Dae-jong said the short-term effect on the semiconductor cycle is limited. He identified whether new leaders will keep Nvidia’s approach to technology development and investment as a key issue. The report also said Nvidia’s plans after its Feynman roadmap, which extend beyond 2028, depend on the new leadership’s strategic consistency.
Nvidia CEO exit unsettles AI stocks and lifts the dollar
Asian markets steadied on Friday after Nvidia CEO Jensen Huang resigned a day earlier, shaking AI-linked stocks and helping push the dollar higher as Treasury yields rose. Investors were also watching tighter U.S. chip export limits, weaker copper demand and a growing chance of a 50-basis-point Federal Reserve rate hike. The article says concern about the risks of AI projects could lead investors and lenders to demand higher financing costs for data centers. Traders also reassessed whether AI construction would create strong near-term demand for copper and other industrial metals.
Trading firms weigh the risks of moving from AI copilots to autonomous systems
A session at Finance Magnates Summit London 2026 will examine how trading firms can move AI tools from demonstrations into trusted operations. Speakers will discuss data quality, permissions, accountability, governance and how firms can track and control an AI agent’s decisions. They will also consider where brokers should build their own tools and where outside providers may help. The session is scheduled for 10:10 to 10:50 on 24 November and will cover execution analytics, automation and the skills businesses may need as routine work changes.
Nvidia trades at 10.9 times 2027 earnings, second lowest among AI giants
Nvidia trades at 10.9 times estimated calendar 2027 earnings, the second lowest figure among the AI companies listed, after Micron at 4.3 times. The figures are Broadcom at 17.2 times, TSMC at 18.2, Alphabet at 19.9, Microsoft at 23.7 and AMD at 33.2. Nvidia closed Thursday at $166.19, down from $173.73 on Wednesday.
A hybrid AI model learns when deeper reasoning is worth the cost
Researchers introduce a hybrid decision model that either answers in one forward pass or uses full step by step reasoning when the added value is worth the cost. A lightweight routing block estimates whether deeper reasoning will improve the answer, and a price threshold determines when to use it. For agent tasks without labels for each decision, the researchers developed Calibrated Decision Policy Optimisation, which trains the router using episode level results. They tested the method on MMLU-Pro, AGIEval, GPQA-Diamond, JevBench, BANKING77 and MiniWoB. On JevBench, HDM-4B scored 0.857 compared with Jev’s 0.851, with a median time of 0.179 seconds versus 0.652; on MiniWoB, HDM-27B reduced token cost by 80.1 percent.
Microsoft outage raises questions about staffing and AI investment
Microsoft has offered buyouts as part of efforts to reduce its workforce while investing billions in AI infrastructure. Earlier this year, the company projected $10 billion in AI investments. There is no evidence that the workforce cuts or AI spending directly caused Thursday’s outage. The disruption has raised broader concerns about whether companies reducing staff while reorganizing around AI may have fewer resources to maintain and troubleshoot existing systems.
Different definitions of AI safety may complicate US China talks
Analysts say US China AI safety talks face a challenge over how the two countries define serious incidents. Washington is concerned about losing control of AI systems, while Beijing gives priority to stability and content rules, experts said. The discussions come ahead of the Apec summit. The article also reports that the two governments have established a new dialogue mechanism, as they share an interest in preventing AI related harm.
Anthropic’s Dario Amodei Faces Growing Pushback From Tech Rivals
Anthropic CEO Dario Amodei has angered fellow technology leaders with warnings about the dangers of powerful AI models. Anthropic and OpenAI urged the White House to limit the development and sharing of some models, while Microsoft CEO Satya Nadella and then Nvidia CEO Jensen Huang backed keeping them open. Two dozen companies quickly signed a letter supporting open models, followed by 200 more, including OpenAI; Amodei refused to sign. Tensions have grown as Anthropic takes the lead in AI development, reports strong revenue growth and moves toward a possible major public offering, while critics also object to its business practices and efforts against free alternatives.
Sources
- AMD vs. Marvell: One AI Stock Won the Rebound, the Other Is the Buy, Says Investor
- Nvidia turmoil reshuffles the AI trade. Winners, losers and what investors should watch
- Could AI reduce shipping costs and grocery bills?
- Daol Securities warns Nvidia faces roadmap and AI investment uncertainty after Huang exits
- Nvidia’s CEO Exit Rattled AI Stocks And Lifted The Dollar
- AI & Trading 2027: From Copilots to Colleagues
- Nvidia trades at 10.9 times 2027 earnings, cheaper than every AI giant on this chart except Micron
- The Value of Thinking: Learning When to Reason in Hybrid Decision Models
- Microsoft's massive outage puts its AI-driven layoffs under scrutiny
- Why differing views on safety could cloud US-China AI dialogue
- Why Anthropic’s Dario Amodei Has Angered OpenAI and Other A.I. Peers
Comments
Please log in to post a comment.