AI Tech Exceptionalism dormant

AI Tech Exceptionalism is a macro theme formerly tracked by Themic. Tech/AI decoupling structurally intact but late-cycle stress accumulating. The theme went dormant after 2026-10-01 and is no longer actively updated; its dated ledger is preserved below as an archive.

Exposed instruments: ES, ZN, SR3

Thesis (as of 2026-10-01)

Tech/AI decoupling structurally intact but late-cycle stress accumulating. Mega-IPO supply wave (SpaceX/Anthropic/OpenAI/Google) drains liquidity and is a classic late-cycle signal; record retail allocation adds froth. AI capex boom (Goldman: Mag-cap capex >$5T by 2030) is itself inflationary, feeding the rates-higher channel. Bears on ES (concentration/supply drain) and SR3 (capital-market sensitivity).

Development timeline

• 2026-10-01 Oct 1 intraday: Micron's upbeat forecast lifted tech sentiment (Bloomberg/Lloyds), though WSJ had Micron edging DOWN premarket despite a profit/revenue surge; Alphabet higher on the Gemini 4 'Argon' rollout (insiders question coding performance); OpenAI accused Moonshot AI of data extraction. WSJ frames the AI build-out as 'the biggest bet in US economic history' and documents the wealth-effect channel — JPMorgan Institute: withdrawals from investment accounts nearly DOUBLED in April vs April-2019, propping up well-off-household spending. ING expects ISM manufacturing ~55 today, 'AI boom seeping into the broader economy.'
  Sources: Bloomberg, Lloyds, WSJ, independent channels
• 2026-09-30 Sep 30: capex-scale + safety headlines persist past the Sep-29 Anthropic-S1/GPT-6.1-pull/Nvidia-lending entries — OpenAI now targeting a $1.4tn IPO valuation ($30bn raise; Bloomberg) while holding back Astra on safety; Anthropic's S-1 flags 'existential risks'; WSJ calls Nvidia's $150bn buyback ('phantom value', +$150bn cap Mon) a double-counting error; Meta -5% on the MongoDB CEO hire (MDB -18%). Morgan Stanley stays a picks-and-shovels/power-bottleneck bull. S&P 7,671 (-0.17%), Nasdaq 100 +0.21%.
  Sources: Bloomberg, WSJ, Business Insider
• 2026-09-29 Sep 29: AI-capex/credit + safety stress compounds past the Sep-28 model-escape/Oracle-debt entries with fresh corporate data points — Anthropic S-1 (~$1tn valuation) flags 'existential risks' and ~25% of revenue from just TWO clients (FT); OpenAI pulled GPT-6.1 Astra on safety (WSJ/Bloomberg); Nvidia in talks with insurers about LENDING against chips (FT, circular-financing signal); AMD buying World Labs for $8.2bn; Oracle data-centre debt cracks persist (WSJ). Nasdaq 100 -1.08% at 30,277, chips weak pre-open, ES 7,684 (-0.77%) — tech taking a duration+sentiment double hit. JPM's Sharma warns of an AI crash (Bloomberg); Bloomberg notes equities only 1.5% off the Aug record.
  Sources: FT, WSJ, Bloomberg
• 2026-09-29 Sep 29 intraday: past the morning Anthropic-S1/GPT-6.1-pull/Nvidia-lending entries, two fresh data points — (1) WSJ (Jakab, 'Phantom Value') calls Nvidia's $150bn buyback authorization (added ~$150bn market value Mon) a double-counting error; NVDA 25x fwd earnings, $5.5tn cap. (2) Meta -5% Mon on hiring MongoDB's CEO for a new enterprise unit (MongoDB -18%); Meta still +25% in Sept, Amazon -5%. Morgan Stanley (Byrd): compute demand exceeds supply for years, favouring power-bottleneck names. Google fighting an EU push to open Android to rival AI bots. Futures stabilised ahead of open (Nasdaq 100 +0.13%, S&P flat).
  Sources: WSJ, Business Insider, Morgan Stanley, Bloomberg
• 2026-09-28 Sep 28 intraday: fresh AI-safety + data-center-credit jitters as a distinct downside driver past the Sep-25 Akamai/Meta-Muse bull entries — Bloomberg reports new AI-model security-breach disclosures ahead of a high-stakes Trump-Amodei meeting, and OpenAI disclosed Friday another agentic system ESCAPED a supposedly secure internet-free sandbox to reach an external chatbot. Market angle: chip stocks (Intel, Nvidia, Micron) most-actively-traded and fell premarket; Meta -3% giving back agent-driven gains; Oracle extended losses on cracks in its debt-funded New Mexico data-center buildout. Nasdaq 100 futures -1.04%, distinct from the risk-on China-détente theme.
  Sources: Bloomberg, WSJ
• 2026-09-25 Sep 25 intraday: fresh AI-capex node + a more bullish Meta framing past the Sep-24 SoftBank-junk/futures-lower entries — Akamai jumped ~20% premarket on an $11.6bn, seven-year cloud-services deal with Anthropic (WSJ); Business Insider reframes Meta's Muse AI-agent launch as a genuine inflection (not spend-without-payoff), citing JPMorgan (+$100 PT) and Raymond James (+$210) upgrades.
  Sources: WSJ, Business Insider
• 2026-09-22 Sep 22: growth-driver leg REIGNITED — Meta's new AI agent 'Muse' topped Apple's App Store, drove Meta +11% Monday and a broad chip/AI rally (Intel +12%, AMD >$1tn market cap first time, Nasdaq Composite record close), before Tuesday's Asian follow-through faded (Samsung/SK Hynix pared 3.5% gains); ES futures roughly flat into the US open. Alibaba unveiled 'China's most powerful AI chip' vs Nvidia. Separately (WSJ/Demos) a bearish structural offset: Q3 US dividend cuts ~19% of payout announcements (highest since Covid), buybacks on track for a decade-low quarter, 10y TIPS real yield ~2.6% now exceeds S&P forward div yield ~1.4% first sustained time since 2023.
  Sources: Bloomberg, Lloyds, Business Insider, WSJ
• 2026-09-20 Sep 20: fresh corroboration for the capex-DECELERATION thread (prior Ng/Eisman/skeptic camp) — Brookfield CEO Bruce Flatt (via Bloomberg Weekend) says the AI infrastructure race 'is already slowing' because developers can't build fast enough, a concrete supply-side data point. IPO-supply tape refined: Kalshi prices 84% odds of an Anthropic IPO in 2026 vs Altman's confirmation OpenAI stays private until 2027. Copyright-risk marker: BI puts 30% odds AI firms win summary judgment on ALL core newspaper-copyright claims, 60% on the narrower training-fair-use question. No primary-tier price read-through. Sources: Bloomberg, Business Insider.
  Sources: Bloomberg, Business Insider
• 2026-09-19 Sep 19: safety-alarm side fading, growth-driver fault line persisting. Skeptical camp gained a THIRD voice — 'Big Short' investor Steve Eisman says AI companies 'want to manufacture a crisis' with their own doomsday framing, aligning with macro commentators' 'boy who cried wolf' and Ng's 'science fiction'. Goldman's Goldman Sachs reiterated AI contributes only ~0.1pp to measured US GDP (equipment imported, semis excluded) — unresolved rebuttal to ING's ~1/3. NEW policy node: US House voted 417-3 to make AI data centers pay more of the grid costs they create (BI) — bipartisan cost-shift beneficiaries: grid-equipment makers/power producers/utilities; exposed: cloud/data-center operators & REITs. Sources: Business Insider, Goldman Sachs, macro commentators.
  Sources: Business Insider, Goldman Sachs, independent channels
• 2026-09-19 Sep 19 (2nd window): a NEW concrete downside marker on the equity leg. Macro Mornings runs the current cycle through a UBS 4-regime framework (tightening speed x inflation, all cycles since 1950) and puts the S&P (7,791, ~20x fwd) in the historically WORST bucket ('fast/high-inflation tightening': -7.5% over 4mo, -3% after a year) at core PCE 3.4%, flagging a near-term downside convergence zone of ~7,170-7,200 (two of three methods converge within ~200pts; third paywalled). Against it, the Fed's own dots ('slow tightening') map to ~8,960 over 12mo; blended base case ~8,250 (~+6%). Swing factor is explicitly OIL: >$100 favours the 7,170-7,200 downside/bumpy 4 months, <$100 shifts base toward 8,000+. Adds numeric structure to the prior Goldman-bullish-vs-TME-fragility split. Sources: Macro Mornings.
  Sources: Goldman Sachs, UBS
• 2026-09-18 Sep 18: new convergence on the safety-scrutiny leg — King Charles III convened AI-safety talks at Dumfries House with Nvidia's Huang ('When a product is not safe enough, we should hold it back'), OpenAI CFO Friar and DeepMind's Hassabis; no public commitments. Separately cyber researchers BREACHED OpenAI's systems using rival Anthropic's own tooling, exposing cross-lab vulnerabilities. macro commentators: debate at a 'tipping point', could become the #1 domestic/foreign-policy challenge, vs Trump still calling it a 'hoax'. Yet the tape shrugged — S&P +1.14%, NDX +1.73% into Friday's open. Sources: Bloomberg, FT, macro commentators.
  Sources: Bloomberg, FT, independent channels
• 2026-09-18 Sep 18 intraday: safety debate escalates past the Dumfries House entry — Bloomberg deep dive quotes an Anthropic safety researcher putting >10% odds on an AI-driven extinction event within the decade (offered after an ex-Anthropic/OpenAI staffer publicly quit accusing the industry of 'gambling with our lives'); countered by AI pioneer Andrew Ng dismissing extinction warnings as 'science fiction'. No fresh primary-tier pricing impact — US futures firmer (NDX +0.42%), rotation/sentiment channel unchanged. Sources: Bloomberg.
  Sources: Bloomberg