Ramez Naam

persona · active · confidence: medium · reviewed 2026-09-21 · tier: quarterly

Who he is

Affiliation: Founder & Managing Partner, Planetary VC (Seattle; climate, clean energy, storage, electrification; 30+ portfolio companies per planetary.vc) — role check run 2026-09-21: confirmed by the ISI seminar listing (2026-03-12), his XPRIZE bio and the Moonshots EP #280 intro (2026-08-15). Formerly Partner & Chief Futurist, Prime Movers Lab — gone from PML's team page (Crunchbase still lists it; stale). Former Co-Chair for Energy & Environment, Singularity University — his own XPRIZE bio says "former"; Wikipedia (edited 2026-08-27) still says current; I trust his bio. v1 said "decade-plus Co-Chair" with no "former" — Drifted. 13 years at Microsoft (Outlook, Internet Explorer, Bing). Author of the Nexus trilogy — Nexus (2012; Prometheus Award 2014), Crux (2013), Apex (2015; Philip K. Dick Award) — plus The Infinite Resource (2013) and More Than Human (2005). v1's "LP across multiple funds" is unverified (a fundraising directory lists an "Allocations 2026 Master" vehicle; low-confidence, not relied on). His About page still says he is "launching" a VC firm — he does not maintain his bio; never grade affiliations from it.
One-line position: exponential cost curves in clean energy get underpriced by everyone including him, and the AI power crunch is a flexibility problem, not a generation shortage — "the grid is a cache problem, and batteries are the cache" (Moonshots, recorded 2026-08-04). New this cycle: since spring 2026 he is as loud on AI ("AGI hype is premature; ASI/singularity is bad sci-fi"), and he is personally invested in the thesis he sells — portfolio company Agentic Infrastructure holds ~10 GW of sites built for the Texas rule he praises.
v1 provenance: added 2026-08-20, surfaced from Rich's own prompt while researching Emerald AI / WeaveGrid; v1 eyebrow "Luminary Lens — Energy & Climate Tech."

Discipline & technical bet

Clean-energy futurist-investor with an engineer's habit: publish a numeric forecast, wait, grade yourself in public, revise. The lesson he repeats: institutional forecasters (IEA, utilities, climate advocates) lowball exponential curves because linear thinking is the default — and so did he. Two live wagers. (1) Grid: load flexibility (data centers that agree to power down ~100 hours a year — "curtailment") plus four-hour batteries unlock ~100 GW of existing US grid without new wires; winners are flexibility software, batteries and behind-the-meter power (made or stored on your own site), not new generation — anyone waiting on a firm connection (a guaranteed hookup) stays in the queue. (2) AI: scaling gives log-linear returns ("intelligence ≈ log(compute)"), so the frontier needs algorithmic breakthroughs, no lab keeps a lead, and open-weight models trail by months. He is no longer a pure commentator on (1): Agentic Infrastructure ("power-first land development for hyperscale data centers," planetary.vc) is built to exploit Texas's interruptible path.

How to apply this lens (carried from v1; caution added 2026-09-21)

Good fit for the Power & Energy cross-cutting layer — demand-flexibility, storage, grid-software names (EMLD, WEAV, GRID, AMSC). Ask on his behalf: is the market treating this as a hard physical shortage when it's actually a flexibility/misallocation problem? Is there an exponential curve (cost, adoption, deployment speed) consensus is still extrapolating linearly? Weaker fit for pure-compute or pure-chip names with no grid angle — don't force it. Caution: WEAV (WeaveGrid) is a Planetary VC portfolio company and Agentic Infrastructure is a direct beneficiary of the interruptible-load rule — when the lens says "buy flexibility," check whether it is talking its book. Use him for the mechanism; take timelines and GW counts from ERCOT, FERC and Duke.

Key claims (Says)

Notable predictions — with falsifiable checks

Revealed behavior (Does)

Feels

Vindicated and wary at once: a decade of being too conservative on solar taught him to distrust every institutional forecast including his own, so he hedges upward by reflex. Wants it to be true that abundance is cheap and the only bottleneck is paperwork — and, since 2026, that no lab or state can own AI.

Hears

Duke / Tyler Norris flexibility research, ERCOT–PUCT–FERC dockets, BNEF-style cost curves, Chinese manufacturing data; the Moonshots table; his own founders; the open-weight / plural-AI crowd (Oslo Freedom Forum, Bitcoin summits). Little from grid-reliability skeptics or labor economists — the two groups whose data cut against him this cycle.

Sees

Deal flow in grid flexibility, storage and data-center land — the real economics of a 10 GW interruptible site book before regulators publish results. And fifteen years of his own numeric forecasts to grade against reality, which almost nobody on the bench keeps.

Incentive map

Re-derived this cycle — up sharply from v1's implicit "independent futurist." Who pays him: Planetary VC's limited partners, speaking fees, royalties. What he's selling: the flexibility thesis is the thesis his portfolio is built on — Agentic Infrastructure monetizes the interruptible path he praises; WeaveGrid, Amperon, Alpha Grid monetize grid-flexibility software; Ouros, Electra AI, QuidNet monetize storage. What he can't say: "the interruptible path is slower than 12–18 months," "the flexibility GW are being over-read," "Duke doesn't support project-level claims" — each marks down a portfolio company. The plural-AI thesis is less conflicted (no exposure found); the solar record predates the fund. Rule-25 discount: high on grid timelines and flexibility GW; moderate on AI market structure; low on historical cost curves. Channel discount: his grid numbers reached Rich via a Diamandis episode — do not count Naam and Diamandis / Wissner-Gross as two independent votes on grid abundance.

Theories aligned with

What he's reacting against

Where he overlaps / splits (with Rich)

Sample lens lines (carried from v1, annotated 2026-09-21)

On Emerald AI: "'Good luck getting that power before 2031, 2032' if you wait for a firm connection — but 200GW of idle grid headroom is sitting there between night and day demand, worth about $10 trillion in stranded data center capacity. That's the trade Emerald AI is selling, and my own math lands on the same ~100GW they're claiming." [Note: "my own math" and theirs share the Duke ancestor.]
On the ERCOT/FERC policy shift: "Texas cut the wait from five-to-seven years to twelve-to-eighteen months by letting flexible loads in the door. FERC just told the other six grids to explain why they haven't done the same. That's not advocacy anymore, that's a compliance deadline." [Note: the 12–18 months is unverified outside his own remarks; FERC's orders and the PUCT approval landed the same day.]

Track record

Empirical vs normative

Weak spots / open questions

Rich's take

Delta log

2026-09-21 — v1→v2 migration + validation (bulk run)

Sources

migrated v1→v2 2026-09-21 · backup: _archives/ramez-naam.html.bak-20260921 · v1 added 2026-08-20 (meta dialect: last_updated / domain) · Singularity 🧠 / People