The 7 Taxes on AI

Key Takeaways

• TSMC Q1 revenue $35.9B, up 40.6% YoY.
• ASML Q1 revenue $9.6B, exceeds expectations.
• AI model wars matter less than infrastructure 'tax' providers.
• Seven AI taxes span silicon, networking, cooling, interconnection, custom silicon.
• 2026 AI capex exceeds $690B, boosting tax companies.

Pulse Analysis

The latest earnings from TSMC and ASML illustrate how AI‑related demand is reshaping traditional semiconductor and equipment markets. TSMC’s 40% revenue jump and ASML’s €8.77 billion (≈$9.6 billion) top‑line underscore a surge in high‑performance computing chips, especially for AI accelerators that now account for more than half of TSMC’s sales. While analysts chased headline‑grabbing model competition, the underlying infrastructure that enables inference—fabrication plants, extreme‑ultraviolet lithography, and chip‑design tools—has become the true growth engine, insulated from short‑term hype.

Investors should therefore shift focus from the speculative race among OpenAI, Anthropic, and Google to the seven "taxes" on AI identified in the thesis. These include TSMC’s fabs, ASML’s EUV machines, EDA firms that design chips, high‑speed networking providers, power‑and‑cooling specialists, data‑center interconnection operators, and the bespoke silicon packaging partner that services multiple hyperscalers. Each passes rigorous tests of structural necessity, pricing power that rises with demand, regulatory or physical moats, and revenue orthogonal to the model war. As AI capex climbs past $690 billion in 2026—driven by Amazon’s $200 billion, Google’s $180 billion, Meta’s $125 billion, and Microsoft’s $150 billion commitments—these firms are positioned to capture a toll on every dollar spent.

The timing is critical. Macro noise from geopolitical tensions and policy debates temporarily depresses valuations, creating entry points for disciplined investors. Unlike pure‑play AI model stocks whose margins may compress as customers build alternatives, the tax companies enjoy compounding pricing power and limited competition. By anchoring portfolios to these infrastructure pillars, investors can gain exposure to the long‑term AI spend trajectory while mitigating the volatility tied to model‑centric narratives.