The Download: open model trade-offs and geothermal power’s promise

The newsletter highlights debate over open AI models after Anthropic warned that Z.ai’s freely downloadable model could raise cyberattack risks, alongside Meta’s launch of new AI models. Fervo Energy plans to expand enhanced geothermal power with a $2.2 billion public offering and an expanded Google agreement; its Utah facility is also reported online as the world’s first utility-scale enhanced geothermal plant. Other items include a cyberattack that overheated refrigerators at hundreds of French supermarkets and a case raising questions about legal protections for brain-implant users.

Analysis

Investment read-through: The clearest investable signal is not that AI demand makes enhanced geothermal attractive, but that power procurement is becoming a potential constraint on AI infrastructure growth. Fervo Energy (FRVO) could benefit if its Utah plant demonstrates repeatable output and project economics; the expanded Google (GOOG) agreement is a demand signal, not proof that geothermal can be deployed at scale or at competitive cost. For Google, the strategic value is diversified firm power, but one project is unlikely to change consolidated earnings near term. Fervo’s financing and build-out also make execution, capital needs, and dilution central risks.

Meta (META) model progress may improve product engagement and strengthen its position in the developer ecosystem, but benchmark performance alone does not establish durable monetization. More capable, freely available models can also commoditize model access and raise safety, legal, and infrastructure costs across the sector. Watch product adoption and spending evidence, not capability claims.

The supermarket incident is a reminder that cyber risk can cross from IT into physical operations, creating potential demand for industrial/OT security and resilience. The article does not identify a listed beneficiary or establish a sector-wide earnings impact. The image-matching tool raises privacy and platform-governance questions for Airbnb (ABNB) and Zillow (Z), but the likely transmission is regulatory or reputational and remains unquantified.

Timing / contrarian view: Near-term headlines may reward AI capability and power scarcity; over 1–3 months, operating and financing disclosures matter more. The market may be underweight the difficulty of scaling geothermal, while over-reading early model scores. Thesis-breakers include weak plant performance or adverse project economics, financing terms that materially dilute FRVO holders, and evidence that AI usage does not translate into sustained product adoption.

Key Decisions for Investors

• Keep FRVO on a catalyst watchlist; do not chase the AI-power narrative before verifying plant output, project economics, financing structure, and the terms and scale of its Google agreement. Reassess over the next 1–3 months as disclosures arrive.
• For GOOG, treat the agreement as strategic optionality rather than a near-term earnings catalyst. Track whether firm-power procurement broadens and whether it affects infrastructure costs or capacity plans.
• For META, avoid trading benchmark headlines alone. Revisit the thesis if model adoption, product engagement, or monetization improves without a corresponding deterioration in spending or risk disclosures.
• Monitor OT cybersecurity demand after the refrigeration incident, but make no direct equity trade on this report alone: there is no named listed supplier or evidence of a measurable sector revenue impact.