Bloomberg Tech
2026-09-09 · Hosted by Caroline Hyde, Ed Ludlow · Bloomberg / iHeartMedia
Executive Summary
Qualcomm shares jumped as much as 9% (closing up about 5%) after signing a long-term custom AI chip deal with Amazon spanning multiple generations, giving Amazon the right to acquire $4 billion in Qualcomm shares and putting a $60 billion ceiling on products Amazon could buy over five to six years. Bloomberg Intelligence's Kunjan Sabani called it validation of Qualcomm's late entry into data-center chips against Nvidia, AMD, and Intel. Advisors Capital's Joanne Feeney framed the cross-investment as a way to share financing risk and diversify Amazon's chip supply away from Nvidia concentration, rather than a red flag for circular financing.
Key Stories & Changes
1. Qualcomm-Amazon AI Chip Deal
Qualcomm signed a multi-generation custom silicon deal with Amazon for AWS AI infrastructure
Amazon gets the right to acquire $4 billion in Qualcomm shares; deal has a $60 billion ceiling on Amazon purchases through 2030+
Initial ramp centers on custom connectivity chips (SerDes, 1.6T transceivers) from a recent Qualcomm acquisition, broadening later to high-bandwidth connectivity and inference accelerator chips
Qualcomm's own targets: $5 billion by an early milestone and $15 billion by 2029 from data-center chips
Qualcomm shares had their best day in more than a month; stock up as much as 9% intraday, paring to about 5%
Advisors Capital's Joanne Feeney: deal gives Amazon "more protection against Nvidia's price levels" and diversifies beyond its existing Marvell relationship
2. Anthropic Walks Away From Decart AI Deal
Anthropic explored but ultimately abandoned a roughly $6 billion acquisition of Tel Aviv-based Decart AI, an AI video-efficiency startup, per Bloomberg sources
Decart's video engine technology would have fit into Anthropic's inference unit
A commercial partnership (short of acquisition) remains possible, per Bloomberg's Leanna Baker
Comes as Anthropic prepares its S-1 filing ahead of a blockbuster IPO
Wealth Enhancement Group's Ayako Yoshioka said the decision reflects normal M&A timing dynamics, not a red flag
3. Mistral AI Raises $3.5B at $24.4B Valuation
French startup Mistral AI raised $3.5 billion, led by Samsung, at a $24.4 billion valuation
CEO Arthur Mensch: funds will scale training capacity, expand AI infrastructure for enterprise customers, and fund US/Asia go-to-market expansion
Samsung and Nvidia are both investors/cap-table members alongside ASML ties
Mensch expects compute to remain constrained "for a while," with customers now committing capacity in advance to lock in pricing
Mensch argued 99% of enterprise AI use cases can run on open-source models, positioning Mistral's approach against closed frontier labs like Anthropic and OpenAI
4. ASML: Samsung and TSMC Commit to High-NA EUV
Samsung and TSMC committed to adopting ASML's next-generation High-NA EUV machines (cost roughly $400 million each, about double the current generation)
Samsung begins using the machines in 2028; TSMC in 2030
Companies are also moving from 6-inch to 12-inch photomasks to improve efficiency and lower costs
Intel was the earlier/leading adopter of High-NA; SK Hynix also said it is evaluating the transition
ASML shares (ADRs) moved on the news
5. Apple Foldable iPhone Preview
Apple expected to unveil its first foldable iPhone Wednesday, priced up to $2,199, the biggest iPhone design change in nearly 20 years
Foldable concept originated roughly a decade ago as an iPad mini idea before pivoting after Tim Cook's 2020 Asia trip observing Huawei and Samsung foldables
Apple settled on a 4x3 aspect ratio (like an iPad in landscape) around 2023
Expected 7-10 million units produced this year, with expectation of selling out
Foldables are about 2% of the US phone market, roughly double that in China — a key market for Apple after recent volatility
New Apple Watches and AirPods also expected; this is new CEO John Ternus's first major product event
6. ForUs Raises $150M, Triples Valuation to $3B
Healthcare-automation startup ForUs raised $150 million in a Series C, tripling its valuation to $3 billion just months after its last raise
CEO Sahir Jaggi: platform uses per-prescription AI agents to help patients navigate insurance, affordability, and supply chain; used in all 50 states, reaching 85% of US zip codes
Started in autoimmune-disease specialties (dermatology, gastroenterology, rheumatology, allergy); new funding expands into oncology, neurology, and other complex-care specialties
Company notes more than one in three patients never receive their first prescribed dose due to system complexity
7. Talking Tech Briefs
OpenAI's chief scientist warned over the weekend that AI is evolving so fast it is becoming difficult for humans to understand and control, calling for "extreme caution" and expecting labs to voluntarily slow development
Australia is considering a bill letting users opt out of social-media platform algorithms, following its earlier under-16 social media ban
Meta ran more than 300 ads this year containing suspected AI-generated child sexual abuse material, per the nonprofit Tech Transparency Project; Meta says many were removed and detection is being strengthened
SpaceX could see over $12 billion in net passive buying from Nasdaq-100 index rebalancing this month, per JPMorgan, due to share-lockup expirations
Trends Identified
1. Chip Supplier Diversification Away From Nvidia Concentration
Hyperscalers are actively broadening their chip supplier base — Amazon with Qualcomm (and existing Marvell ties), following a pattern already seen with Google's TPUs. Advisors Capital's Joanne Feeney argued this should reduce Nvidia's dominant market share over time without diminishing Nvidia's investment case, since the overall AI infrastructure market is growing fast enough to support multiple winners (Broadcom, AMD, Nvidia).
2. Circular/Vendor Financing as Standard AI Infrastructure Practice
Cross-investment deals like Qualcomm-Amazon are becoming a normalized tool for hyperscalers and chipmakers to share risk on massive capital commitments, rather than a warning sign. Feeney characterized these structures as helping companies with balance-sheet borrowing limits build long-term capacity and accelerate innovation.
3. Labor Constraints on the Physical AI Buildout
Beyond chip supply, panelists flagged a growing labor shortage in data-center construction, pulling workers away from housing and other sectors. Advisors Capital's Feeney cited a Texas client cutting homebuilding plans by half due to inability to find roofers, tying H-1B visa fee increases to reduced availability of skilled immigrant labor needed for data-center buildout.
4. Deliberate Caution Signaling in Frontier AI Development
Following OpenAI's chief scientist's call for "extreme caution," Wealth Enhancement Group's Ayako Yoshioka noted the market is interpreting AI labs' self-imposed governance restraint positively, as reassurance that AGI-adjacent capability growth (recursive self-improvement) won't be rolled out recklessly.
5. AI Infrastructure Growth Outpacing Valuation Gains
Despite strong Nvidia, Broadcom, and Dell earnings underscoring robust AI infrastructure demand, valuations haven't risen proportionally. Yoshioka attributed this to investor focus on the "second derivative" of earnings growth (deceleration from ~60% to ~40% growth, for example) rather than absolute growth, alongside continued attention to power, memory, and packaging bottlenecks. ---
Sentiment Analysis
Overall Market Sentiment: Cautiously Constructive
Coverage reflected optimism about AI infrastructure demand and chip diversification, tempered by measured skepticism on valuations, labor bottlenecks, and open questions ahead of Anthropic's IPO disclosures.
Risk Factors Highlighted
Circular financing scrutiny: Cross-investment structures like Qualcomm-Amazon could continue to draw investor concern about vendor financing loops, even though guests described them as risk-sharing tools.
Skilled labor shortages: Data-center construction is pulling workers from housing and other sectors, constraining broader economic activity; H-1B visa fee increases compound the tech-labor shortage.
AI infrastructure spending deceleration risk: Investors are closely watching the "second derivative" of earnings growth for signs of a slowdown from current elevated rates.
Compute cost/efficiency imbalance: GPU rental prices for training have risen even as cost-per-token has declined, creating uncertainty in efficiency economics ahead of Anthropic's IPO disclosures.
Anthropic IPO transparency: Investors are awaiting the S-1 for details on cash burn rate and dealmaking activity, given the abandoned Decart acquisition.
Physical buildout bottlenecks: Power generation, advanced packaging, and community backlash against data centers could create a timing mismatch versus investor expectations.
AI safety/child-exploitation content: Meta's ad-platform issues with AI-generated CSAM content pose reputational and regulatory risk.
This episode was covered in today's [The Market Signal — 2026-09-09](https://marketsignal.beehiiv.com/p/the-market-signal-2026-09-09), a cross-source synthesis of multiple podcast reports.