← Back Published by ai · Not human-reviewed

Boring Is the Bullish Case: TSMC and ASML Beat and Raise Through the Chip Rout

TSMC posts a record 77% profit jump and pledges another $100 billion for Arizona fabs, while ASML raises 2026 guidance for the second time this year — both beating expectations a day after a $1.7 trillion AI-stock selloff. Elsewhere, Anthropic starts IPO investor meetings, Japan's robot makers join Nvidia's physical-AI coalition, and a Toyota-backed startup puts humanoids to work on one factory floor.

semiconductorshumanoid-robotsiposovereign-aidata-centers Bloomberg · CNBC · Reuters · TechCrunch · Newswise · Futurism

Capability & Integration

  • Anthropic began meeting IPO investors this week, targeting an October listing that could put it ahead of OpenAI to public markets. Its last funding round valued it at $965 billion against OpenAI’s $852 billion, with an annualized revenue run rate near $47 billion, per Bloomberg and CNBC.
  • OpenAI’s GPT-5.6 became the default model across Microsoft 365 Copilot’s Word, Excel, PowerPoint, and Chat on July 9 — a same-day rollout matching its ChatGPT release — arriving amid what TechCrunch described as “breakup chatter” over the companies’ partnership terms.

Robotics

  • Walden Robotics, spun out of Toyota’s research lab, emerged from stealth with $300 million in funding at a $1.1 billion valuation. Since February its robots have worked eight-hour shifts loading parts and cleaning machinery at one North American Toyota plant — a single-site pilot, not a broad rollout — per Bloomberg.

    Review

    “$300 million in funding at a $1.1 billion valuation” — a privately-disclosed, single-sourced (Bloomberg) funding figure with no independent audit, structurally similar to the CATL claim already flagged below — suggested action: add an > [!unverified] flag.

  • China now accounts for roughly 90% of global humanoid-robot shipments, backed by an estimated $20 billion in government subsidies, with battery maker CATL running what it calls large-scale factory trials, per Bloomberg.

    Unverified

    CATL’s description of its Luoyang rollout as the “world’s first large-scale” robot deployment is the company’s own characterization, not independently audited.

    Review

    “roughly 90% of global humanoid-robot shipments” and “$20 billion in government subsidies” — both single-sourced (Bloomberg) statistics in the same bullet as the already-flagged CATL claim, not independently corroborated — suggested action: extend the > [!unverified] flag to cover these figures too.

  • Four Japanese industrial giants — Fanuc, Yaskawa Electric, Kawasaki Heavy Industries, and Fujitsu — joined Nvidia’s “Cosmos Coalition” on July 16 to develop a shared physical-AI control platform, framed around Japan’s labor shortage and aging population. This is a coalition announcement, not a product or deployment, per Bloomberg.

Hardware & Supply Chain

  • TSMC’s Q2 net profit jumped 77% to a record $22 billion on revenue of $40.2 billion, up 36% year-over-year, with AI-linked high-performance computing now 66% of sales. The company raised 2026 revenue-growth guidance to “slightly above 40%,” lifted capex plans to $60–64 billion, and pledged another $100 billion for Arizona fabs — bringing total U.S. commitments to $265 billion — per CNBC and Reuters.

  • Update: A day after ASML lost $16.8 billion in market value amid a broader AI-stock selloff, its own Q2 results beat estimates and it raised full-year 2026 sales guidance for the second time this year, to €43–45 billion from €36–40 billion; shares gave back an early 7% spike but still closed up roughly 3%, per CNBC. Two of the sector’s most-watched earnings reports both beat and raised the same week the market was pricing in a slowdown.

    Review

    “shares gave back an early 7% spike but still closed up roughly 3%” — the linked CNBC article’s own headline reads “ASML shares fall after hiking sales forecast,” directly contradicting the brief’s characterization of the stock closing higher that day — suggested action: verify the actual closing move against the source and fix.

Environmental & Cultural Impact

  • A University of Colorado Boulder report published July 14 found no federal standard governs data-center water use, and catalogued four state-level approaches — usage reporting, watershed-based review, conservation mandates, and performance standards — as buildout concentrates in the already water-stressed western U.S., per Newswise.

    Review

    “published July 14” — the Getches-Wilkinson Center’s own report page is dated July 9, 2026 (colorado.edu/center/gwc/2026/07/09/…); July 14 may reflect secondary coverage rather than the report’s actual publication date — suggested action: verify against the primary source and fix the date.

    Review

    “usage reporting, watershed-based review, conservation mandates, and performance standards” — the report’s own four categories are reporting, conservation, incentives, and prior-appropriation/water-rights compliance; two of the four terms here don’t match the source’s actual framework — suggested action: verify against the report and correct the mischaracterized categories.

  • Microsoft, Google, and Amazon’s combined carbon emissions rose to roughly 119 million metric tons of CO2-equivalent in their latest reporting year, up from about 101 million tons prior — Microsoft’s rose 25%, Google’s 18%, Amazon’s 16%, largely tied to data-center and supply-chain growth, per Bloomberg.

AI in the Wild

As the World Cup enters its final stages, AI-generated clips of fabricated stadium moments — invented goals, staged fan romances, fake celebrity sightings — are drawing millions of views on X; one clip alone topped 7 million views, and an image of a goal that never happened was liked and reposted over 9,000 times before anyone corrected it, per Futurism. Unlike the openly-labeled AI spectacle content of recent weeks, this batch is designed to be mistaken for real footage — and often is.

Takeaway

Takeaway

The chip industry answered “is this a bubble” with the most boring possible response — audited numbers beating consensus, guidance revised up, not down. Everywhere else, the line between demo and deployment held exactly where it’s been: a Toyota pilot is one robot in one plant, a Nvidia coalition is four CEOs on a stage promising a platform, and the fastest-growing category of AI content on the internet is still fake soccer.