Scan: Will the US start the process of building a nuclear-powered data center on a military base before 2030?

date	predicted	market price	confidence
2026-08-29	60%	54%	44%
2026-08-22	53%	53%	50%
2026-06-24	66%	54%	50%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: kalshi_direct kalshi_related polymarket_related claude_news article_search gdelt_news congress_bills wikipedia

Sub-questions (Fermi decomposition)

What is the current Kalshi price and volume on KXDATACENTER-30, and how has it trended?
What specific actions count as 'starting the process' under Kalshi's rules — does the May 2025 executive order directing DoD to designate military sites for nuclear-powered AI data centers, or an RFP/site selection, already qualify?
Has the Department of Defense (Army Janus program, Air Force, DIU Advanced Nuclear Power for Installations) awarded any contracts or selected sites for microreactors explicitly tied to data centers on military bases?
What is the status of the Air Force Eielson AFB microreactor pilot and other on-base nuclear projects, and are any paired with data center or AI compute workloads?
Have any companies (Oklo, NANO Nuclear, X-energy, Valar Atomics, Radiant, Westinghouse) announced agreements with DoD for nuclear-powered data centers on installations, and what are their timelines relative to 2030?
Has Congress passed or advanced legislation (NDAA provisions) mandating or funding nuclear-powered data centers on military installations?
What do related Kalshi/Polymarket markets on SMR deployment, DoD nuclear reactors, or AI data center milestones imply about timelines before 2030?

Planner reasoning

This is a long-horizon Kalshi market whose resolution hinges on the vague phrase 'start the process of building' — so the key uncertainties are (1) how Kalshi will interpret it (does an executive order, DoD RFP, site selection, or Army/Air Force contract award count?) and (2) whether existing initiatives like the May 2025 Trump executive orders directing DoD to site nuclear reactors and AI data centers on military installations, the Army's Janus program, DIU's ANPI, and Air Force microreactor projects (Eielson) already constitute or will soon constitute qualifying action. The market price is the primary anchor; news and cross-market signals will help calibrate how far along these programs are.

2. Research Parallel tool calls

tool	status	time	items	summary
kalshi_direct	OK	0.2s	1	## This Market's Kalshi Data Before 2030 (KXDATACENTER-30) - Current price (probability): 54.00% - 7-day price change: +1.00% - 30-day price change: +4.00% - Average daily volume: 40 contracts - Price range: 45.00% - 63.00% - Data points: 47 days
kalshi_related	OK	2.9s	2	2 related markets / summaries. series KXDATACENTER: 0 markets (skipped 1 no-signal) \| keyword 'nuclear data center': ok \| keyword 'military base data center': ok \| keyword 'microreactor': no matches
polymarket_related	OK	2.9s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'nuclear data center': 0 markets \| keyword 'microreactor': 0 markets \| keyword 'SMR': 0 markets \| keyword 'military base AI': 0 markets
claude_news	OK	19.0s	12	Based on my research, here are the key findings on US progress toward nuclear-powered data centers on military bases: - Executive Order foundation (May 23, 2025): Trump signed four nuclear-related EOs, including EO 14299 "Deploying Advanced Nuclear Reactor Technologies for National Security," w
article_search	OK	3.3s	34	Fetched 34 unique articles across 5 queries (lookback=180d). 'nuclear powered data center military base': 10 articles \| 'Department of Defense microreactor data center': 10 articles \| 'Army Janus program microreactor installation': 10 articles \| 'executive order nuclear reactors AI data centers mili
gdelt_news	OK	161.0s	10	GDELT: 10 articles across 3 queries (lookback=90d). 'nuclear data center military base': error GDELT rate-limited after retries (429) \| 'DoD microreactor contract': error GDELT rate-limited after retries (429) \| 'Army Janus microreactor': 10 hits
congress_bills	OK	3.8s	9	Found 9 bills matching 'nuclear data center military installation'.
wikipedia	OK	2.9s	4	Fetched 4 Wikipedia entries (0 missing pages).

3. Evidence Brief Sonnet 5 · 7744 chars

# Current state As of the research cutoff (~Aug 2026), the US Army's Janus Program has moved from executive order (May 2025) through site selection (9 candidate bases, Nov 2025) to signed contracts (Aug 2026) awarding up to $2.2B to five vendors (Antares/Fort Bragg, BWXT/Fort Campbell, General Atomics/Fort Hood, Radiant/Fort Benning, Westinghouse/Fort Drum). The Air Force's parallel ANPI program has also named preferred sites (Buckley SFB, Malmstrom AFB). Kalshi YES trades at 54%, below where the underlying facts arguably justify given contracts are already signed — the main ambiguity is whether these reactors are explicitly "for data centers" versus general installation resilience/AI-adjacent messaging. # Timeline of key events - 2015: DIU founded (background). (confirmed, Wikipedia) - 2024-06: DIU/Army/Air Force launch Advanced Nuclear Power for Installations (ANPI) program to deploy on-site microreactors at select bases. (confirmed, claude_news) - 2025-04: DIU qualifies 8 vendors (Antares, BWXT, General Atomics, Kairos, Oklo, Radiant, Westinghouse, X-Energy) for ANPI. (confirmed, DIU/datacenterfrontier) - 2025-05-23: Trump signs EO 14299 directing Army to build a domestic military-base nuclear reactor operational by Sept 30, 2028; explicitly ties advanced computing/AI power needs to national security. (confirmed, claude_news, CBS Colorado quoting EO text) - 2025-10-14: Army formally brands effort "Janus Program," reiterating 2028 operational deadline. (confirmed, army.mil) - 2025-11: Army identifies 9 candidate installations; DIU issues Area of Interest solicitation. (confirmed, claude_news, Lexology) - 2025-12-18: FY2026 NDAA (S.1071) becomes Public Law 119-60 (contains nuclear/energy provisions relevant to installations, per congress_bills; specific data-center language not confirmed in text). (confirmed enactment date, content partially unconfirmed) - 2026-04-11: Buckley SFB (CO) and Malmstrom AFB named preferred Air Force ANPI sites; officials estimate completion by 2030, explicitly linking to AI compute needs. (reported, CBS Colorado) - 2026-08-26/27: Army awards ~$2.2B in contracts to 5 vendors for 5 specific bases (Bragg, Campbell, Hood, Benning, Drum) — concrete construction-track step. (confirmed, multiple outlets: CBS, ENR, World Nuclear News, army.mil) - 2026-08-29: Coverage confirms BWXT/other vendor stock reactions to contract award. (confirmed, InsiderMonkey) - Ongoing 2026: FY2027 NDAA (S.4784) in Senate process; no confirmed passage yet. (reported) # Event Will the US start the process of building a nuclear-powered data center on a military base before 2030? # Outcomes to forecast Yes / No # Kalshi market anchor KXDATACENTER-30 YES = **54%** (current). 7-day change +1%, 30-day change +4%; range 45–63% over 47 days of data; avg daily volume only ~40 contracts (thin/low-conviction market). Trend is modestly upward but market remains far from pricing in near-certainty despite concrete contract news. # Sub-question answers 1. **Kalshi price/trend** — 54% YES, +4% over 30 days, thin volume (~40 contracts/day), no strongly directional momentum despite major news flow. [kalshi_direct] 2. **What counts as "starting the process"** — Rules text not provided (none), but EO 14299, DIU site solicitations, and now signed vendor contracts ($2.2B, Aug 2026) plausibly satisfy "starting the process" under any reasonable reading; ambiguity remains on "nuclear-powered data center" vs. general base power resilience framing. [claude_news] 3. **DoD contracts/site selection for microreactors tied to data centers** — Yes: Army Janus awarded contracts to 5 vendors for 5 bases (Bragg, Campbell, Hood, Benning, Drum), Aug 2026, up to $2.2B combined; explicitly tied to AI/computing power resilience in program messaging. [CBS, ENR, world-nuclear-news, army.mil] 4. **Eielson AFB / other pilots paired with data center workloads** — Not directly confirmed for Eielson specifically in this research; Air Force ANPI names Buckley SFB and Malmstrom AFB as preferred sites, explicitly citing AI computing infrastructure needs, targeting completion by 2030. [CBS Colorado] 5. **Company agreements with DoD** — Antares (Fort Bragg), BWXT (Fort Campbell), General Atomics (Fort Hood), Radiant (Fort Benning), Westinghouse (Fort Drum) all under signed Janus contracts (Aug 2026); Oklo, NANO Nuclear, X-Energy, Kairos remain qualified vendors from the original 8 but not yet awarded site-specific contracts per this research. Delivery timelines: reactors expected operational within ~5-year funding window (early-to-mid 2030s), but construction/contracting starts well before 2030. [claude_news, army.mil] 6. **Congressional legislation** — FY2026 NDAA (S.1071) enacted as PL 119-60 (Dec 2025); FY2027 NDAA (S.4784) still in Senate process; specific mandate language for "nuclear-powered data centers" not confirmed in text, only bill titles/dates available. [congress_bills] 7. **Related markets** — No Polymarket matches found. Kalshi's own fusion market (KXFUSION-30) trades at 33%, unrelated but shows lower-confidence framing for hard-tech-by-2030 questions generally; no direct arbitrage market exists. [kalshi_related, polymarket_related] # Key facts (high-confidence, factual) 1. [CBS/army.mil] Army awarded ~$2.2B in Janus contracts to 5 vendors for 5 named bases, Aug 2026 — concrete construction-track commitment. 2. [claude_news] EO 14299 (May 2025) mandates Army reactor operational by Sept 30, 2028 — well before 2030 close date. 3. [CBS Colorado] Air Force names Buckley SFB and Malmstrom AFB, citing AI compute power needs explicitly, target completion ~2030. 4. [DIU] 8 vendors pre-qualified April 2025 under ANPI program. 5. [congress_bills] NDAA FY2026 signed into law Dec 2025; FY2027 NDAA in process. # Cross-market signals - Kalshi related: KXFUSION-30 (nuclear fusion by 2030) at 33% — different technology, limited read-through. - Polymarket: No matching markets found. - Sportsbook: N/A (not applicable to this event type). # Analyst opinions and speculation - Breaking Defense (Oct 2025): nuclear power "coming to an Army base near you, sooner than you think" — bullish framing. - Program messaging blends "AI computing resilience" with general base energy security; unclear if Kalshi resolution requires explicit "data center" labeling vs. general installation power. This is the single largest source of ambiguity/risk for the market. # Directional lean per outcome - **Yes**: Signed $2.2B contracts, five named bases, hard 2028 EO deadline, Air Force sites explicitly citing AI compute needs — strong factual basis that the process has already started well before 2030. - **No**: Risk hinges on strict interpretation requiring explicit "data center" (not just "installation resilience") pairing; if Kalshi resolution requires a literal co-located AI data center project (not just base power), some ambiguity remains, though Air Force ANPI messaging directly cites AI infrastructure. # Gaps / unknowns - Exact Kalshi resolution criteria/rules text not provided ("none" in prompt) — critical ambiguity on what threshold "starting the process" requires. - No confirmed contract or site explicitly labeled as co-located "AI data center" (vs. general base power) at reactor-vendor level. - FY2027 NDAA final content/passage unconfirmed. # Calibration anchors - Kalshi current YES price: 54% (anchor). - Precedent: Given signed contracts, named sites, and a legally mandated 2028 deadline already in force, similar "process already underway" facts in other Kalshi infrastructure markets typically trade 70-85%+; current 54% appears to underprice confirmed contractual/legal facts, though resolution-definition ambiguity justifies a discount below full certainty.

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

52% conf

Yes	66%
No	34%

Kalshi anchors YES at 54% on thin volume (~40 contracts/day). The brief documents a near-deterministic pipeline: EO 14299 mandating an Army base reactor operational by Sept 2028, the Janus Program's ~$2.2B contract awards to five vendors at five named bases (Aug 2026), and Air Force ANPI preferred sites (Buckley SFB, Malmstrom AFB) explicitly citing AI compute needs with ~2030 completion targets. With ~3.5 years remaining and multiple parallel programs, 'starting the process' of a nuclear-powered data center on a base looks more likely than not. The main discount is resolution ambiguity — whether Kalshi requires an explicitly labeled co-located AI data center versus general installation power resilience — plus the usual risk that some of the Aug 2026 contract details are less definitive than reported, which keeps me from pushing much past the mid-60s.

gpt-5.6-sol

69% conf

Yes	76%
No	24%

The 54% Kalshi YES price is a thin-volume anchor, but it appears low given the process has advanced through an executive order, site selection, and $2.2 billion in signed contracts covering five named Army bases. The 2028 operational deadline and Air Force plans for Buckley and Malmstrom, which explicitly cite AI-computing needs, further favor Yes before 2030. Vendor qualification, enacted FY2026 defense legislation, and site-specific agreements reinforce that this is beyond a speculative proposal, while the unrelated 33% fusion market provides little contrary evidence. I retain substantial No probability because no project is conclusively labeled a co-located nuclear-powered data center, and the absent resolution rules leave room for a strict interpretation requiring more than a reactor intended partly for computing resilience.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasters converge on the resolution-ambiguity discount as the dominant driver but neither quantifies how a "reasonable" market resolver would treat "process already started" language given that contracts are signed—this is a key omission since Kalshi questions about processes ("will X start building...") typically resolve YES on credible construction/contracting steps, not completion, which should push probability higher than both estimates.

Neither forecast explicitly addresses that the question asks about "the process of building," not a completed or even explicitly-labeled "nuclear-powered data center"—given EO 14299 explicitly ties compute needs to the mandate and $2.2B contracts are already signed, this is arguably already satisfied today, making 66-76% still oddly conservative if the bar is process-initiation rather than definitional purity.

Both forecasters treat the Kalshi price (54%) as underpricing the facts but don't explore why the market might be rationally low—e.g., traders may believe "data center" requires a literal co-located compute facility (not just base resilience power), a stricter reading that neither forecaster fully stress-tests as the modal resolution path rather than a tail risk.

Neither critiques the reliability of the "thin volume" framing as evidence of mispricing—40 contracts/day could also reflect informed traders correctly pricing high genuine ambiguity, not neglect; both forecasters use "thin market therefore mispriced upward" as an unexamined assumption.

Neither forecast weighs the risk of political/programmatic reversal (funding delays, contract cancellations, changed administration priorities) between now and 2030 despite the 2028 EO deadline already having slipped in analogous defense-tech programs historically—this tail risk of stalling is under-discussed relative to the confidence expressed.

Forecast 2's higher confidence (0.69) versus Forecast 1 (0.52) is not clearly justified by any additional evidence cited—both use nearly identical facts, so the second forecaster's tighter conviction and higher YES probability (76% vs 66%) reflects unexplained overconfidence rather than new information.

6. Aggregator Fable 5.1 · final calibrated forecast

68% (60% confidence)

Yes: 68% No: 32%

The factual pipeline is strong: EO 14299 explicitly ties the Army base reactor mandate to AI/compute needs, $2.2B in Janus contracts are signed for five named bases, and Air Force ANPI sites (Buckley, Malmstrom) explicitly cite AI computing infrastructure with ~2030 targets—so under a 'process started' reading, resolution to Yes is likely and there remain ~3.5 years for an explicit reactor-plus-data-center pairing to be announced. The critique correctly notes, however, that the market has already absorbed the Aug 2026 contract news and still sits at 54%, which suggests traders view the modal resolution path as requiring a literal co-located data center rather than general installation resilience power; that is a real risk, not a tail. Program slippage and political reversal add modest downside, though they matter less for a 'start the process' question than for completion. I land between the two forecasts, above the market anchor but discounting for the unresolved definitional ambiguity and the market's failure to react more strongly to concrete contracting.

Pipeline Timing

Total pipeline time: 259.3s

Per-tool research timings shown in the Research section above.