Research Frontier: Energy

What's actively moving in 2026. This file is the discovery input for the content skill — keep it current and specific.

---

Active Frontiers

1. SMR / Advanced-Reactor Commercialization — 2026 Is the Regulatory Gate Year

Status: Rapid progress — NRC decisions imminent; hyperscaler demand now underwriting scale-up Key players: NuScale Power, TerraPower, Oklo, X-energy Key sources: nuscale-q1-2026-context (SEC 8-K), meta-66gw-nuclear-deal (Latitude Media), hyperscaler-nuclear-procurement

The SMR story has two layers in 2026: the regulatory gate and demand-side underwriting. On regulation, NuScale remains the only SMR with full NRC design certification (standard design approval granted Jan 2023) — its 77 MWe module scales to 924 MWe (12 modules) in a VOYGR plant, and that certification is the reference point against which Oklo (first reactor targeted ~2027) and others are measured (nuscale-q1-2026-context). TerraPower's Natrium completed its safety evaluation in Dec 2025 but still lacks an NRC construction permit. On demand, hyperscalers are now directly underwriting the commercialization curve: Meta's up-to-6.6 GW deal commits 1.2 GW to Oklo (first reactors as early as 2030) and 2.6 GW + 1.2 GW storage to TerraPower (two Natrium units as soon as 2032) (meta-66gw-nuclear-deal).

What to watch in the next 90 days:

• TerraPower Natrium NRC construction permit decision (expected H1 2026)
• X-energy Xe-100 NRC safety evaluation (November 2026 target)
• HALEU domestic supply agreements — the critical missing link for Natrium and Xe-100
• Re-ingest NuScale's Q1 2026 8-K full-text (revenue, net loss, cash, backlog still unread — fetch was HTTP-403 blocked)

Open problems:

• First-of-kind SMR cost estimates consistently exceed projections (NuScale UAMPS $5.3B → $9.3B precedent)
• Can a 15 MWe Oklo Aurora design reconcile with a 1.2 GW Meta target? (Many units, or an undisclosed larger product)
• Neither Oklo nor TerraPower holds an NRC construction permit yet — can they deliver on 2030/2032 dates?

---

2. Hyperscaler Nuclear PPAs — A Single Buyer Class Underwrites the Curve

Status: Rapid progress — the defining near-term energy story Key players: Constellation Energy, Vistra, Talen Energy, Oklo, TerraPower Key sources: hyperscaler-nuclear-procurement, meta-66gw-nuclear-deal (Latitude Media)

Hyperscalers are procuring nuclear three ways to meet AI data-center load: (1) restart existing plants, (2) first corporate SMR PPAs, and (3) multi-GW advanced-reactor development deals (hyperscaler-nuclear-procurement). The anchor deals trace the arc: Constellation↔Microsoft (Sep 2023) to restart Three Mile Island Unit 1; Google↔Kairos Power (Aug 2025), the first corporate SMR PPA (Hermes 2 / Oak Ridge, 50 MW by 2030 scaling to 500 MW); and Meta↔Vistra/Oklo/TerraPower (2026), up to 6.6 GW by 2035 for the "Prometheus" supercluster in New Albany, Ohio — a scale Latitude Media likens to all of northern Virginia's "Data Center Alley." The gating constraint on AI buildout is power, and nuclear PPAs are how hyperscalers secure firm, carbon-free baseload outside congested grid-interconnect queues.

What to watch:

• NRC restart license decision for Crane Clean Energy Center / TMI (expected mid-2026)
• Whether additional hyperscalers replicate the multi-GW advanced-reactor model (Meta is the template)
• FERC ruling on behind-the-meter co-location (Talen/Amazon model) — industry-defining

Open problems:

• What price is being paid? (Most nuclear PPAs are confidential; estimates $80–120/MWh)
• If SMR construction slips, do the Google/Meta SMR + advanced-reactor contracts trigger termination clauses?
• How is Meta's Vistra leg framed — "2.2 GW Ohio + 433 MW upgrades" (Latitude) vs "~2,600 MW PJM" (Vistra earnings)? Same commitment, or additive?

---

3. Fusion Milestones — Private Capital Consolidates Around Hardware Proof

Status: Early-stage → rapid milestone progress Key players: Helion Energy, Commonwealth Fusion Systems Key sources: helion-polaris-150m-milestone (Fortune)

On 2026-02-13, Helion's 7th-gen prototype Polaris (suburban Seattle) reached 150 million °C (~10× the sun's core) and was reported as "the first privately developed fusion machine to operate with deuterium-tritium fuel" (helion-polaris-150m-milestone). The ~100M°C minimum threshold for sustained commercial fusion was cleared. But the milestone is bounded by hard caveats: as of late 2025 Polaris had not announced net energy production, and reactor assembly for the first commercial Orion plant (50 MW, Malaga, WA) had not begun. The Microsoft PPA targets first grid electrons in 2028 — widely seen as aggressive. Helion's future target fuel is deuterium + helium-3. CFS (SPARC; Gates + Nvidia backed) is the principal competitor.

What to watch:

• Any Helion announcement of net energy production from Polaris (the gating proof; absent as of late 2025)
• Orion reactor assembly start (not begun as of late 2025) — the leading indicator for the 2028 PPA
• CFS SPARC first plasma (2027 target) and its 200 MW Google ARC PPA

Open problems:

• The 2028 Microsoft timeline requires net energy + commercial scale in ~2 years from a machine not yet assembled
• D-He3 fusion (Helion's long-term fuel goal) is further from demonstration than D-T
• Commercial fusion LCOE remains completely unknown — no credible cost models published

---

4. GE Vernova's Equipment Backlog — The Gas Bridge Is Huge

Status: Rapid progress — record order intake Key entities: GE Vernova

GE Vernova is the clearest near-term beneficiary of the AI power cycle. Its gas turbine backlog reached 100 GW in Q1 2026, with the total company backlog at $163B (+$13B in Q1 alone) — projected to reach $200B by 2027, one year ahead of prior forecast. The Electrification segment booked $2.4B in datacenter-related orders in Q1 2026 alone — more than all of 2025. The constraint is manufacturing capacity, not demand; transformer lead times of 2-5 years mean equipment ordered today ships in 2027-2029.

What to watch:

• Q2 2026 GEV earnings (expected July 2026): Does backlog keep growing?
• GEV manufacturing capacity expansion announcements
• BWRX-300 SMR: GEV/GE Hitachi Canadian deployment progress

---

5. Uranium Market — Supply Discipline Meets Demand Acceleration

Status: Active — price recovering after brief retreat Key entities: Cameco, Kazatomprom

Uranium spot hit $101/lb in January 2026 before pulling back to $83-85/lb. Kazatomprom cut 2026 output 10%. Cameco signed a C$2.6B long-term supply deal with India. Long-term contracting is accelerating as utilities and nuclear PPA-backed projects lock in multi-decade fuel supply.

What to watch:

• Spot price recovery: Does uranium return to $100+ in H2 2026?
• HALEU enrichment: Does DOE fund Centrus expansion or new capacity?
• India / South Korea / EU mega-deal contracting

---

Recent Breakthroughs

Earlier dated developments (NIF ignition 2022, GE Vernova Q1 2026 backlog, uranium $100/lb, FERC CIR waiver, etc.) are tracked in timeline.md.

---

Predictions & Trends

• 2027 is the nuclear inflection year — Crane (TMI) restarts for Microsoft; Natrium construction permit issued and underway; nuclear PPAs begin delivering electrons
• A single buyer class now sets the SMR curve — Meta's 6.6 GW deal proves hyperscalers will underwrite pre-permit advanced reactors (Oklo, TerraPower) on 2030/2032 dates
• SMR first commercial power is not before 2030 — construction permit ≠ online power; realistic window 2030-2033 for demonstration plants
• Fusion proof bar moves to "net energy" — Helion's 150M°C clears temperature; the next credibility gate is net energy production and Orion assembly, both absent as of late 2025
• GE Vernova $200B backlog by 2027 — multi-year contracted revenue; equipment supply tightness is the near-term constraint on AI datacenter buildout

---

Knowledge Gaps

Areas needing more primary sources:

• NuScale Q1 2026 financials — the 8-K was fetch-blocked (HTTP 403); revenue, net loss, cash & equivalents, backlog, and latest module-order count are unread. Re-ingest via authenticated path. (nuscale-q1-2026-context)
• Meta 6.6 GW deal — primary documents — figures are from a single news source (Latitude Media); corroborate via Meta / Vistra / Oklo / TerraPower filings and PPA terms.
• Hyperscaler-procurement aggregation — the synthesis file's original Build.inc URL is dead (HTTP 404); the Sep-2023 / Aug-2025 PPA dates and SMR figures want a primary-source pass.
• Helion net energy & Orion assembly status — Fortune reports neither as achieved by late 2025; track for the first hard confirmation.
• Transformer supply chain — manufacturers, lead-time data, pricing, geographic concentration (Korean/European dependence) for AI datacenters
• Kairos Power KP-FHR — technical details and Hermes 2 construction timeline less documented than Natrium/Xe-100
• Palisades restart — Holtec progress; DOE loan terms; precedent for the first-ever US reactor restart after closure
• European nuclear — EDF, Fortum, CEZ AI-driven demand and deal structure
• Long-duration storage vs nuclear — at what LCOS does iron-air / flow battery storage displace nuclear for 24/7 CFE?