Bloomberg Tech
2026-09-03 · Hosted by Caroline Hyde, Ed Ludlow · Bloomberg / iHeartMedia
Executive Summary
Nvidia is reportedly closing in on its biggest AI acquisition yet: a roughly $14 billion deal to buy Hugging Face, the open-source AI model hub, including about $1 billion earmarked for staff retention. Bloomberg's Ian King and Rachel Metz, who broke the story, said the deal appears imminent and could be announced this week; Nvidia wants broader adoption of AI across the economy and sees Hugging Face's open-model ecosystem as a way to expand its customer base beyond the hyperscalers.
Key Stories & Changes
1. Nvidia Nears $14 Billion Hugging Face Acquisition
Deal size of ~$14 billion includes roughly $1 billion in staff retention payments; up from an earlier reported figure of $12.9 billion
Would be Nvidia's largest-ever AI acquisition; the company has historically preferred in-house development and investment stakes over acquisitions (prior notable deal: assets of Grok/xAI-adjacent chip investor)
Hugging Face operates a platform for sharing and showcasing open-source AI models; recently was the subject of a security incident involving OpenAI models
Nvidia has reportedly been a long-standing investor in Hugging Face; deal reflects alignment on open-model strategy
Reporters say the deal is "pretty imminent," expected to be finalized this week; no contract signed yet
2. Dell Raises AI Server Outlook by $25 Billion
Annual sales outlook raised by $25 billion; AI server revenue expected to triple to $74 billion this fiscal year
$95 billion AI server backlog; earnings beat Street expectations broadly
Fifth consecutive quarter of raised guidance; traditional (non-AI) server revenue up 110% year-over-year
Stock up 252% year-to-date; operating margins improved to over 11% from 6% last year, reflecting a product-line optimization program
PwC report cited: global data center spending could reach $32 trillion by 2050, up sharply from a prior 2030 forecast of $8 trillion
DELL: Dell Technologies — Shares higher, off session highs — Raised AI server outlook by $25B; record backlog
SNOW: Snowflake — Reporting after the bell — Revenue expected to rise ~30% for a third straight quarter
HPE: Hewlett Packard Enterprise — Reporting after the bell — Stock up ~112% in 2026 on AI server demand
AVGO: Broadcom — Reporting after the bell — Stock down ~$520B from June highs; investors want fiscal 2027 AI outlook clarity
3. Uber Cuts 10% of Global Workforce
About 3,300 roles, roughly 10% of global staff, aimed at cutting management layers
20% fewer managers expected; CEO Dara Khosrowshahi cited internal survey feedback about excess coordination overhead
Company reaffirmed investment in core ride-hailing/delivery and international expansion (competing with DoorDash in Europe)
Uber committed over $10 billion to autonomous-vehicle strategy over the coming years, partnering with Nuro, Lucid, Rivian, and Avride rather than building its own robotaxis
4. Cognition Raises ~$1 Billion at $47 Billion Valuation
New valuation of $47 billion, up sharply from $26 billion in May
Company received nearly $10 billion in total investor interest for the round
Annualized revenue passed $900 million, up from $492 million in May
SpaceX had previously approached Cognition about an acquisition; Cognition declined, preferring to remain independent (SpaceX separately closed a $60 billion acquisition of Cursor in August)
5. Alphabet Avoids Forced Ad-Tech Sale
A federal judge ruled Google does not have to sell its advertising exchange, instead requiring interoperability with rival ad tech tools
Stock move was muted; ruling follows an April 2025 decision that found Google illegally monopolized two ad-tech markets
6. Napster Reinvents as Agentic AI Company via Microsoft Azure
Napster CEO John Acunto detailed the brand's pivot from music streaming (shut down in January) to an "agentic AI" interface layer built natively on Microsoft Azure
Claims cost advantage of about a penny a minute versus 25 cents to a dollar for competitors
Positions Q4 2026 into 2027 as the "launching pad" for customer adoption
7. Humanoid Robotics: Barclays Sees $200 Billion Market by 2035
Barclays' Zhinita Torozova forecasts the humanoid robotics market could reach $200 billion by 2035, part of a broader $3 trillion physical AI market
60,000 new humanoid robots expected to deploy globally in 2026 (up from 15,000 in 2025); 85-90% of current deployments are in China
Could reach 13 million new humanoids deployed annually by 2035, with up to 30 million in operation if current momentum holds
Biggest bottleneck: lack of real-world physical training data, not demand; simulation seen as a partial but insufficient solution
Trends Identified
1. AI Infrastructure Spending Is Still Accelerating, Not Slowing
Dell's $25 billion outlook raise, its $95 billion backlog, and PwC's eye-popping $32 trillion 2050 data-center forecast all point the same direction: infrastructure demand shows no signs of plateauing. Guests connected this directly to Nvidia's own recent guidance of 70% supply-driven growth, suggesting the constraint remains capacity, not customer appetite.
2. Nvidia Is Buying Distribution, Not Just Technology
The Hugging Face deal reflects a strategic shift from investment stakes toward outright acquisition to secure broader AI adoption across the economy. By owning the open-model hub, Nvidia positions itself as infrastructure for the entire open-source AI ecosystem, not just the hyperscaler layer, complementing its G20 messaging (from CEO Jensen Huang) urging global AI infrastructure build-out.
3. AI Capital Is Chasing Growth-Stage Coding and Model Startups at Extreme Multiples
Cognition's jump from a $26 billion to $47 billion valuation in roughly four months, alongside nearly $10 billion in investor demand for a $1 billion raise, shows private AI valuations continuing to run well ahead of even fast-growing revenue (annualized revenue nearly doubled in the same window).
4. Custom Silicon and Optics Are Broadening the Semiconductor Story
Guest Michelle Canal flagged that copper interconnects can't keep pace with AI data transmission needs, pointing to optics and new materials as the next investable layer — beyond Nvidia and Broadcom into MediaTek and private semiconductor firms — with the caveat that market-share shifts will create both winners and losers.
5. Humanoid Robotics Is Entering a Scale-Up Phase Bottlenecked by Real-World Data
Barclays' forecast of a 4x jump in annual humanoid deployments this year underscores accelerating momentum, but the constraint has shifted from demand to the availability of physical-world interaction data — an emerging sub-market in itself. ---
Sentiment Analysis
Overall Market Sentiment: Bullish on AI Infrastructure
Coverage skewed strongly positive on AI infrastructure demand (Dell, Nvidia, humanoid robotics), with more mixed-to-cautious framing around labor costs (Uber layoffs) and looming earnings tests (Broadcom, HPE).
Risk Factors Highlighted
Broadcom fiscal 2027 AI revenue uncertainty: Investors want confirmation the company can hit $120-150 billion in AI revenue; last quarter's guide disappointed markets.
Custom silicon competition for Nvidia/Broadcom: Rising competition from MediaTek and other chipmakers could erode market share as the semiconductor stack broadens.
Data center political backlash: G20 commentary, including Treasury Secretary Bessent's remarks that AI companies have "done a horrendous job" explaining themselves to the public, signals growing political risk to the buildout.
Humanoid robotics data bottleneck: Lack of real-world physical interaction data, not demand, is the primary constraint on scaling general-purpose humanoid robots.
Napster/agentic AI credibility: Repeated host skepticism about whether a legacy consumer brand pivot to "agentic AI" reflects substance or marketing.
Labor disruption from AI-driven restructuring: Uber's 10% workforce cut, while framed as efficiency-driven, reflects a broader pattern of AI-era organizational streamlining with real employment impact.
Alphabet regulatory overhang persists: Despite avoiding a forced ad-tech sale, Google still must make ad tools interoperable with rivals, an ongoing compliance requirement.
This episode was covered in today's [The Market Signal — 2026-09-03](https://marketsignal.beehiiv.com/p/the-market-signal-2026-09-03), a cross-source synthesis of multiple podcast reports.