Bloomberg Tech
2026-09-23 · Hosted by Caroline Hyde, Ed Ludlow · Bloomberg / iHeartMedia
Executive Summary
President Trump addressed the UN General Assembly, declaring the US rejects a "global scheme" to control AI and pledging to call the technology "superintelligence" going forward. Markets barely reacted: equities were largely flat, the US 10-year yield sat at 4.95%, and the dollar spot index held steady. The speech set the stage for a Trump-Xi Jinping meeting later this week, where AI is expected to be a central topic alongside a possible framework for US-China AI dialogue.
Key Stories & Changes
1. Trump's UN Speech Rejects Global AI Oversight
President Trump told the UN General Assembly the US rejects attempts to "construct a global scheme" to control AI
Said future US documents will refer to the technology as "superintelligence" and that the US will "encourage" rather than "rein in" its development
Framed AI dominance as central to winning the technology race against China
Market reaction was muted: equities flat, US 10-year yield at 4.95%, Bloomberg dollar spot index little changed
Senator Elizabeth Warren sent a letter to Trump expressing "significant concern" about his "dangerous approach to artificial intelligence" ahead of talks with Xi
2. Trump-Xi Meeting Looms Over AI Policy
Treasury Secretary Scott Bessent met his Chinese counterpart in New York; the two sides have set up a framework for a US-China AI dialogue covering risks, benefits, and a possible notification system
No details confirmed on deliverables or whether China will send a complementary delegation of executives
Nvidia CEO Jensen Huang expected at Thursday's state dinner alongside execs from Microsoft, OpenAI, Apple, and Qualcomm
Nancy Tengler (Laffer Tengler Investments) said she expects "platitudes" rather than a breakthrough, predicting China will not change its chip-export behavior
3. Alibaba's AI Chip and AGI Push
Alibaba unveiled what it calls China's most powerful AI chip, three times more powerful than its prior generation released just three to four months ago
The chip, referred to as the Gemway v900, is an accelerator built for frontier training, aimed at competing with Nvidia and domestic rivals Huawei and Cambricon
Alibaba committed more than $53 billion to AI and plans to build 20 gigawatts of compute capacity
CEO Eddie Wu said the company's primary objective is now artificial general intelligence (AGI), targeting an LLM with five to ten trillion parameters
Alibaba has been "overshadowed" recently by smaller Chinese AI upstarts like DeepSeek and Moonshot
4. AI Agent Momentum: Muse and Grok Bot
Meta's Muse AI agent hit #1 on Apple's free download chart, extending Meta's 11% stock jump from the prior session
Nancy Tengler said she recently added Meta to her portfolio, citing its 20% compounded revenue growth, attractive relative price-to-sales ratio, and Muse as a way to finally monetize CapEx spend (capital expenditures, i.e. money spent building infrastructure)
SpaceX AI's Grok bot topped 400,000 weekly users (418,000 as of September 14), a 24% week-on-week jump, since its August 11 launch
SpaceX AI shares rose about 1.5% on the news
5. Andreessen Horowitz Launches AI-Era Academy
Horowitz and Dreson Academy, a new alternative-education venture backed by $42 million, launches in San Francisco for high school graduates aged 16-20
Backed by OpenAI, Anthropic, and Nvidia founders among others; Mark Andreessen joins the board along with general partner Eric Torrenberg
CEO Gargan Biani said the school prioritizes "pursuits" (80%+ of student time) over traditional majors, focusing on AI-native building, people skills, and ambition
First founding-class fellowship (one year, tuition-free) opens for applications now; program starts September 2027
A16Z's portfolio has roughly 850 companies with 18,000 open roles, underscoring the talent-pipeline rationale
6. World Models and AI Safety with Fei-Fei Li
Dr. Fei-Fei Li, CEO of World Labs, discussed the company's new world model Atlas, which generates 3D/4D pixel-consistent environments rather than words, distinguishing it from LLMs
Use cases span creative industries, design, robotics, and pharmaceutical lab science, where world models can simulate realistic training/evaluation environments
Li said there is currently no sufficient benchmarking standard for world models and called for ecosystem-level responsibility involving academia, government, and industry
She rejected fatalism on existential AI risk, saying the greater threat lies "within ourselves" — in how humans govern and use the technology
7. SoftBank's Junk Bond Sale Draws Strong Demand
SoftBank's junk bond sale has drawn more than $20 billion in early demand
The company seeks to raise over $11 billion (a $10 billion tranche plus a €1 billion tranche) to help fund its OpenAI investment
Pricing expected Thursday
Trends Identified
1. AI Agents Enter Their "Moment"
The last 24 hours have been dominated by discussion of AI agents reaching a tipping point, with Meta's Muse topping app store charts and SpaceX AI's Grok bot posting a 24% weekly user jump. Guests connected this to broader monetization narratives — Nancy Tengler argued Muse finally justifies Meta's AI CapEx spend, marking a shift from skepticism to conviction among some investors.
2. US-China AI Rivalry Intensifies Ahead of Trump-Xi Meeting
Alibaba's new chip announcement, timed just before the Trump-Xi meeting, underscores accelerating competition in AI hardware and frontier model development. Both governments are signaling toughness — Trump rejecting global AI oversight, China pushing its own chip and LLM ambitions — while a nascent bilateral AI dialogue framework offers only a modest de-escalation channel.
3. Divergence Between Industry Safety Calls and US Policy
Frontier lab leaders (Dario Amodei, Sam Altman) continue to publicly call for guardrails, benchmarking, and vetting frameworks, while the Trump administration explicitly prioritizes speed over regulation, calling existential risk concerns a "hoax." This creates friction between industry's stated safety posture and the administration's deregulatory stance, with Congressional pushback (Senator Warren) emerging as a check.
4. AI Education and Talent Pipeline Investment
The launch of the Horowitz and Dreson Academy reflects a broader shift among venture-backed institutions to build alternative talent pipelines tailored to AI-native skills rather than traditional degrees, reinforcing Silicon Valley's view that conventional higher education is lagging the pace of technological change. ---
Sentiment Analysis
Overall Market Sentiment: Cautiously Bullish
Investors and guests broadly framed AI developments positively, though with skepticism toward any near-term geopolitical breakthrough between the US and China.
Risk Factors Highlighted
US-China AI dialogue may yield no substantive agreement: Multiple sources expect the Trump-Xi meeting to produce "platitudes" rather than a formal framework, leaving chip export policy unresolved.
Existential AI risk dismissed by the administration: Trump called existential risk concerns a "hoax," diverging sharply from industry safety advocates, raising governance uncertainty.
Lack of world model benchmarking standards: Fei-Fei Li noted no sufficient ecosystem-level benchmark exists to evaluate safety and capability of world models.
Congressional pushback on AI policy: Senator Warren's letter signals potential domestic political friction over the administration's AI stance heading into the Xi meeting.
Alibaba's capital constraints vs. Western rivals: Peter Elstrom noted Alibaba "doesn't have the same kind of money" as Western AI companies despite its $53 billion commitment.
Weapons delivery delays to allies: Bloomberg reported the US is warning European allies of up to five-year delays in weapons deliveries amid efforts to rebuild domestic industrial capacity — a geopolitical risk tied to broader Trump foreign policy.
Iran escalation risk: Trump threatened to "annihilate" the Islamic Republic absent a deal, an unresolved geopolitical risk that could affect broader market sentiment.
This episode was covered in today's [The Market Signal — 2026-09-23](https://marketsignal.beehiiv.com/p/the-market-signal-2026-09-23), a cross-source synthesis of multiple podcast reports.