NASA

What this page is: the standing page for the agency — a temperature check on space, read through Jared Isaacman's own words, scored against what actually happens. Updated as he speaks and as NASA publishes. The person page holds who he is; this page holds what the agency says, buys and does.

What NASA sets, and what it doesn't

• The Moon is his #1 objective. 17:10: "number one, national space policy objective, return American astronauts to the lunar surface, build the moon base out." What he won't promise is the lunar economy — that regolith / manufacturing returns more than it costs (18:56).
• He does not set the AI/orbital-compute demand. 8:34: NASA is "structurally very disadvantaged relative to any of the hyperscalers", who spend "many times NASA's budget" on hardware. On orbital data centres: "if Elon and SpaceX are betting on this… there is no reason to believe this will not come into existence" (17:10). NASA is a beneficiary at the end of that chain, not its source.
• What NASA does set: its own procurement (SAM.gov / NSPIRES), prizes, and the "near-impossible" lane it wants — nuclear power and propulsion, the "Rickover nuclear navy" move (16:18).
• The demand chain on Jared Isaacman is assembled from several answers, not recited in order.

🌡 Temperature check — his statements, scored

Rule: every new Isaacman statement with a date or number gets a row, and every row is re-verified against current reality before it is quoted anywhere.

🤖 How NASA uses AI — the record (researched 2026-09-16)

The pieces are public; an agency-wide answer to "how will NASA use AI" does not appear to be.

What Isaacman said (EP274)

1. On-mission autonomy — DAVINCI (Venus) designed "from the get-go" to have on-board AI decide what data to send home before the probe dies. 6:49
2. Genesis Mission, theme 1 — "what have we missed": AI over decades of archival mission data, and the coming flood (Roman: 100× Hubble's field of view). 10:10
3. Genesis Mission, theme 2 — "extending our reach", which he narrows to propulsion: beyond chemical, toward fission, fusion, eventually antimatter. 11:05–12:50
4. Frontier-model access (Dave asked about Mythos / Fable 5): he declined the policy question, pointed to OSTP, and named nuclear-programme data as NASA's AI accelerant. 74:38
5. Science throughput: agrees AI + additive manufacturing could turn SMD's $7B/yr into far more missions than one flagship — "a cultural change too." 73:49

What he has said since (on X)

6. 09-12, Space Academy essay: "we have no idea how quickly that timeline could accelerate in a world transformed by artificial superintelligence, robotics, quantum technologies and additive manufacturing." — the Administrator using the word superintelligence in a policy post.
7. 09-15, quoting the President: "the worst choice is to fall behind and hope our adversaries will responsibly wield a technology with the potential to elevate nations and change civilizations. The best choice is to continue to ensure America has the superior AI." 248K impressions, 8.5K likes — the most-engaged thing he posted in two weeks was about AI, not space.

What is actually published

The gap — the story nobody has written

• Theme 1 (mine the archive) is everywhere on paper. Theme 2 (propulsion / reach) is nowhere. The administrator's stated AI priority for going further has no entry in NASA's own AI register. ⚠️ Caveat: the inventory predates the 07-22 Genesis commitments, and R&D may be under-reported in it.
• AI for the #1 objective is arriving through science, not operations. The May register had nothing deployed for Artemis; since then the Lunar Foundation Model (09-10) and ASTRA (09-15) point at the Moon — but both map terrain or scout. Nothing public yet on AI for crew, landers, or base logistics.
• No agency-wide "how NASA uses AI" document found. Science has a strategy; the agency doesn't (publicly).
• Why it matters to GForce: the Moon base leans on humanoids and autonomy (13:46), and NASA's own register shows almost no human-spaceflight AI in deployment. That's where outside teams — and prizes — fit.

📡 Pulse — what people are saying (scan 2026-09-16)

Node: scripts/nasa_pulse.py (X, relevance-sorted, 7 days) + /last30days + scripts/space_rail.py. Raw files in data/research/nasa-ai/. Patterns, not quotes-as-fact — these are opinions.

1. NASA's AI news only travels when a big-tech partner carries it. IBM's Lunar Foundation Model post: ~1,400 engagements. Every NASA-side and reposted version: single digits. NASA's own AI story has no amplifier of its own.
2. Isaacman's AI line outran his space lines. The 09-15 "superior AI" post got 8.5K likes; the Space Academy essay 3.3K; "months, not years" 1.5K. His audience responds to AI-as-national-race more than to NASA programmes.
3. The orbital-data-centre conversation is a stock story ($SPCX), not an engineering one. Top posts are ticker takes ("dominated by only one company"); the sceptics argue cost (~100× terrestrial) and unrepairable hardware depreciation. Polymarket puts an orbital data centre of 100+ accelerators at 2% by end-2026, 21% by end-2027 ($53K volume) — against SpaceX's "2027".
4. Nobody on X is connecting NASA to AI procurement. The solicitation conversation is Rocket Lab's MTO protest, CLPS slipping (SpacePolicyOnline: one CLPS mission left on the 2026 schedule, down from three in May) and LEIA. The AI and the contracts live in separate conversations — the gap is the lane.
5. Isaacman is polarising in procurement circles ("does the bidding of the billionaires" vs "administrator-for-life"). The criticism is about vendor choice, not about AI.

⚠️ Coverage limits: /last30days got no Reddit (403) and no X in this run — only YouTube/GitHub/Polymarket. The X signal comes from our own X API client. Reddit is a blind spot until that's fixed.

🎯 Open NASA calls — from the rail (2026-09-16, 14 days)

How the demand moves (the pattern this fortnight): Isaacman states it on X (Moon Base, nuclear, Space Academy) → NASA posts a sources-sought / RFI → a BAA or RFP with a deadline. LEIA went the whole way in ~4 months (Moon Base announced May → final solicitation 09-08). No AI-specific NASA solicitation is open. AI shows up as open models (IBM) and Genesis commitments, not as something NASA is buying.

Rails — how this page stays current

Related

Jared Isaacman · Moonshots #274 · how to win a nasa challenge · Embodied AI · the timeline