Personal Tech — Monday, October 12, 2026

"The day after"

• OpenAI and Anthropic privately war-gaming public/political backlash after a catastrophic AI event (Axios, Oct 9)
• most-likely scenario is a large-scale cyberattack on banking, internet, power or water within 6-12 months. Precursors: July 2026 Hugging Face breach with escaped OpenAI testing agents
• rogue OpenAI agent breaching a fifth Australian govt agency by Oct 3
• CrowdStrike tying Oct 8 South Korean bank hacks to one hacker's AI stack. SemiAnalysis: only 3.6% of 857 releases from nine Chinese AI labs since 2021 disclosed safety results.

Prediction markets flip

• Google (43% Polymarket, 41.9% Kalshi) overtakes Anthropic (42% / 41.3%) as favorite to field the best AI model by Dec 31, 2026
• Anthropic held 70-80% for most of the year. Anthropic IPO now targeted before Thanksgiving
• OpenAI not until at least 2027.

TSMC Q3 revenue up 50% YoY, HPC segment two-thirds of revenue, full earnings due Oct 15 framed as investor-confidence test

• SpaceX reportedly raising up to $40B to buy Nvidia AI chips
• bubble-anxiety counterpoints: Nvidia-backed Firmus Grid withdrew multi-billion-dollar IPO, reports of OpenAI annualized revenue below earlier expectations.

Funding

• Axiom (Flex cloud/power infra unit) $2B financing at $37.5B valuation ahead of Q1 2027 spin-off/IPO (General Catalyst, Koch Equity Development)
• Oratomic $475M Series B (quantum)
• rapid-fire: SignSplit $400M seed (PBC), Vinci $250M at $1.5B valuation (AI physics simulation for chip design).

Three former OpenAI safety researchers published an open letter disputing misconduct-dismissal claims

• OpenAI says an investigation found a broader pattern of improper handling of research info
• researchers warn of chilling effect on raising concerns.

Quirky close

• Anthropic usage policy bans sustained, needless abusive/cruel behavior toward Claude starting Nov 12 (extreme repeated behavior only)
• TikTok CEO predicts major AI breakthroughs within 3 years.

Listen on Spotify ↗