Morning Edition

The Futurist

Monday, November 09, 2026
AI & Technology Markets & Crypto Ideas Worth Keeping
S&P 5007,376.86▲ 0.27%
Dow Jones51,285.73▼ 0.46%
Nasdaq24,579.43▲ 1.04%
VIX25.64▼ 8.92%
BITCOIN$75,799.00▼ 0.14%
RIPPLE$1.16▼ 0.71%
ETHEREUM$2,232.25▼ 0.38%
DOGECOIN$0.07▼ 1.28%
01

AI & Technology

ALL-IN PODCAST

Midterm Blood Bath, $4T AI Sell Off, Data Center Freeze

A generational GOP collapse in the midterms sent markets reeling, with $4 trillion in AI-related market cap evaporating amid data center freeze fears. The Besties dissect what went wrong for Trump and what comes next for the AI infrastructure buildout.

COINDESK

SpaceX Proposes $24 Billion Texas Supercomputer

SpaceX is pushing a $24B supercomputer project in Texas — worth 17 times its entire bitcoin holdings. The scale signals that private space companies are now serious compute infrastructure players, not just rocket operators.

ALL-IN PODCAST

Anthropic S-1 Shocker, Google's $500B Wipeout

Anthropic's S-1 dropped and the besties are calling it potentially the greatest business ever built. Meanwhile Google shed $500B in market cap in a single day — its worst ever — as AI competitive pressure bites hard.

404 MEDIA

Everyone Hated Meta's New AI Training Terms. Now They're On Pause.

Meta's attempt to tighten its generative AI commercial licensing terms was met with universal backlash and has been shelved. Another reminder that the AI ecosystem's developers hold real leverage over platform terms.

A16Z

Why Frontier Models Still Can't Parse a PDF

Reducto's CEO argues that documents — PDFs, scans, spreadsheets — remain AI's dirty unsolved problem. Until models reliably handle the document layer, agentic AI in healthcare, legal, and finance stays stuck in demo mode.

COINDESK

Solana Moves to Scrap Feature Used in $270M Drift Hack

Solana's developer community is voting to eliminate the transaction mechanism exploited in the massive Drift protocol hack. It's a rare case of a blockchain moving fast to cut a core feature for security reasons.

02

X / Twitter Signal

Trump claims the DOJ was about to jail "a leader of a major AI company" — and Nvidia's Jensen Huang appears to be the subject. $NVDA dropped over 5% in premarket on the news, erasing $225 billion in market cap overnight. Whatever the underlying situation, the signal is clear: AI executives are now operating in a genuinely adversarial political environment post-midterms.

Demis Hassabis amplified his colleague Pushmeet Kohli's observation that DeepMind is experiencing "a genuine phase transition in the pace of scientific discovery." That phrasing — phase transition — is deliberate and precise. It suggests what's happening isn't incremental improvement but a qualitative shift in what AI-assisted science can do.

Ethan Mollick is leaning on Robert Gordon's "Rise and Fall of American Growth" — a pessimistic economic history — to make an optimistic AI case. His point: we've had industrial revolutions before, and "this time is different" has sometimes actually been true. Coming from Mollick, who is rigorous about AI hype, this framing carries weight.

DAIR.AI flagged a sharp warning making the rounds: developers building with AI agents are repeatedly making the mistake of over-relying on someone else's agent harness. As agents get more capable, the scaffolding around them — how they're structured, monitored, and constrained — becomes the actual product. Borrowing someone else's harness means borrowing their assumptions.

03

The Thread

The week's dominant story isn't any single headline — it's the collision of political turbulence and AI infrastructure. A $4 trillion market cap selloff, a potential DOJ action against a sitting AI CEO, and data center freeze fears all hit simultaneously after the midterms. The market is repricing AI not on capability grounds but on political risk grounds, and that's a different and more unpredictable animal.

When political risk becomes the primary variable in AI investing, the roadmap stops being about model benchmarks and starts being about Washington relationships.
The deeper structural story is the document problem. A16z's Reducto interview and the DAIR.AI agent harness warning are pointing at the same gap from different angles: AI agents are becoming capable reasoners wrapped around brittle input pipelines. PDFs, legacy data formats, and poorly structured enterprise data aren't edge cases — they're the majority of real work. Whoever cracks reliable document parsing at scale unlocks the actual enterprise AI opportunity. Meanwhile, SpaceX's $24B Texas supercomputer proposal quietly signals that the compute arms race has moved beyond the hyperscalers. When a rocket company is building data center infrastructure at that scale, the assumption that compute is a Google-Microsoft-Amazon oligopoly deserves revisiting. The physical buildout of AI is becoming its own geopolitical and industrial story — separate from the model wars, and arguably more consequential.