domains/energy/AI datacenter power demand is creating a fusion buyer market before the technology exists with Google and Eni signing PPAs for unbuilt plants using undemonstrated technology.md

@@ -0,0 +1,62 @@
+---
+type: claim
+domain: energy
+description: "Google signed 200MW PPA for ARC (half its output), Eni signed >$1B PPA for remaining capacity, and Microsoft signed PPA with Helion — all contingent on demonstrations that haven't happened yet, signaling that AI power desperation is pulling fusion timelines forward"
+confidence: experimental
+source: "Astra, CFS fusion deep dive April 2026; Google/CFS partnership June 2025, Eni/CFS September 2025, Microsoft/Helion May 2023"
+created: 2026-04-06
+secondary_domains: ["ai-alignment", "space-development"]
+depends_on:
+  - "Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue"
+  - "fusion contributing meaningfully to global electricity is a 2040s event at the earliest because 2026-2030 demonstrations must succeed before capital flows to pilot plants that take another decade to build"
+challenged_by: ["PPAs contingent on Q>1 demonstration carry no financial penalty if fusion fails — they may be cheap option bets by tech companies rather than genuine demand signals; nuclear SMRs and enhanced geothermal may satisfy datacenter power needs before fusion arrives"]
+---
+
+# AI datacenter power demand is creating a fusion buyer market before the technology exists with Google and Eni signing PPAs for unbuilt plants using undemonstrated technology
+
+Something unprecedented is happening in energy markets: major corporations are signing power purchase agreements for electricity from plants that haven't been built, using technology that hasn't been demonstrated to produce net energy. This is not normal utility-scale procurement. This is a demand pull so intense that buyers are pre-committing to unproven technology.
+
+**Confirmed fusion PPAs:**
+
+| Buyer | Seller | Capacity | Terms | Contingency |
+|-------|--------|----------|-------|-------------|
+| Google | CFS (ARC) | 200 MW | Strategic partnership + PPA | Anchored on SPARC achieving Q>1 |
+| Eni | CFS (ARC) | ~200 MW | >$1B PPA | Tied to ARC construction |
+| Microsoft | Helion | Target 50 MW+ | PPA for Polaris successor | Contingent on net energy demo |
+| Google | TAE Technologies | Undisclosed | Strategic partnership | Research-stage |
+
+ARC's full 400 MW output was subscribed before construction began. Google's commitment includes not just the PPA but equity investment (participated in CFS's $863M Series B2) and technical collaboration (DeepMind AI plasma simulation). This is a tech company becoming a fusion investor, customer, and R&D partner simultaneously.
+
+**Why this matters for fusion timelines:**
+
+The traditional fusion funding model was: government funds research → decades of experiments → maybe commercial. The new model is: private capital + corporate PPAs → pressure to demonstrate → commercial deployment driven by buyer demand. The AI datacenter power crisis (estimated 35-45 GW of new US datacenter demand by 2030) creates urgency that government research programs never did.
+
+Google is simultaneously investing in nuclear SMRs (Kairos Power), enhanced geothermal (Fervo Energy), and next-gen solar. The fusion PPAs are part of a portfolio approach — but the scale of commitment signals that these are not token investments.
+
+**The option value framing:** These PPAs cost the buyers very little upfront (terms are contingent on technical milestones). If fusion works, they have locked in clean baseload power at what could be below-market rates. If it doesn't, they lose nothing. From the buyers' perspective, this is a cheap call option. From CFS's perspective, it's demand validation that helps raise additional capital and attracts talent.
+
+## Evidence
+
+- Google 200MW PPA with CFS (June 2025, Google/CFS joint announcement, CFS press release)
+- Eni >$1B PPA with CFS (September 2025, CFS announcement)
+- Microsoft/Helion PPA (May 2023, announced alongside Helion's Series E)
+- Google/TAE Technologies strategic partnership (July 2025, Google announcement)
+- ARC full output subscribed pre-construction (CFS corporate statements)
+- Google invested in CFS Series B2 round ($863M, August 2025)
+- US datacenter power demand projections (DOE, IEA, various industry reports)
+
+## Challenges
+
+The optimistic reading (demand pull accelerating fusion) has a pessimistic twin: these PPAs are cheap options, not firm commitments. No financial penalty if fusion fails to demonstrate net energy. Google and Microsoft are hedging across every clean energy technology — their fusion PPAs don't represent conviction that fusion will work, just insurance that they won't miss out if it does. The real question is whether the demand pull creates enough capital and urgency to compress timelines, or whether it merely creates a bubble of pre-revenue valuation that makes the eventual valley of death deeper if demonstrations disappoint.
+
+Nuclear SMRs (NuScale, X-energy, Kairos) and enhanced geothermal (Fervo, Eavor) are on faster timelines and may satisfy datacenter power needs before fusion arrives, making the PPAs economically irrelevant even if fusion eventually works.
+
+---
+
+Relevant Notes:
+- [[Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue]] — PPAs bridge the gap between demo and revenue
+- [[fusion contributing meaningfully to global electricity is a 2040s event at the earliest because 2026-2030 demonstrations must succeed before capital flows to pilot plants that take another decade to build]] — demand pull may compress this timeline
+- [[the gap between scientific breakeven and engineering breakeven is the central deception in fusion hype because wall-plug efficiency turns Q of 1 into net energy loss]] — PPAs are contingent on Q>1 which is scientific, not engineering breakeven
+
+Topics:
+- energy systems

domains/energy/CFS HTS magnet manufacturing is a platform business that generates revenue from competitors and adjacent industries making CFS profitable regardless of which fusion approach wins.md

@@ -0,0 +1,49 @@
+---
+type: claim
+domain: energy
+description: "CFS sells HTS magnets to Realta Fusion, Type One Energy, and University of Wisconsin WHAM — creating a B2B platform where CFS profits from every fusion approach that uses high-field magnets, plus MRI, particle physics, and industrial applications"
+confidence: experimental
+source: "Astra, CFS fusion deep dive April 2026; TechCrunch April 2026, CFS corporate announcements, IEEE CSC"
+created: 2026-04-06
+secondary_domains: ["manufacturing"]
+depends_on:
+  - "high-temperature superconducting magnets collapse tokamak economics because magnetic confinement scales as B to the fourth power making compact fusion devices viable for the first time"
+  - "Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue"
+challenged_by: ["magnet sales may cannibalize CFS's technical moat by enabling competitors to build fusion devices without developing their own magnet capability; REBCO tape supply chain concentration (top 5 manufacturers control >95% of market) could constrain scaling"]
+---
+
+# CFS HTS magnet manufacturing is a platform business that generates revenue from competitors and adjacent industries making CFS profitable regardless of which fusion approach wins
+
+CFS has pivoted its HTS magnet technology from internal-use-only to a commercial product line, creating three revenue streams: electricity sales from future ARC plants, licensing of proprietary superconducting magnet technology, and manufacturing magnets for external customers. As of April 2026, confirmed magnet customers include:
+
+- **Realta Fusion** — purchasing HTS magnets for their mirror-machine approach (described as "the largest deal of this kind to date for CFS")
+- **University of Wisconsin WHAM** — research-grade magnets for the Wisconsin HTS Axisymmetric Mirror experiment
+- **Type One Energy** — licensed CFS's HTS magnet technology for stellarator reactor design
+
+This is a classic platform strategy: CFS invested $2.86B developing the magnet manufacturing pipeline for SPARC (10,000 km of REBCO tape, 288 toroidal field pancakes, production rate from 30 days/pancake to 1/day). Now they're amortizing that investment across the entire fusion industry. Every fusion startup that uses high-field magnets — tokamaks, stellarators, mirrors — becomes a potential CFS customer.
+
+The manufacturing learning curve is the real moat. CFS's factory has gone through ~6 major manufacturing upgrades. Chief Science Officer Brandon Sorbom: "Our factory now looks a lot more like an auto factory" compared to the early artisanal magnet production. This process knowledge — how to wind REBCO tape into 24-ton D-shaped magnets at production speed — is harder to replicate than the physics.
+
+Beyond fusion, HTS magnets have applications in: next-generation MRI (higher field = higher resolution), particle accelerators (compact muon colliders), maglev transportation, and industrial magnetic separation. Each application expands the addressable market for CFS's manufacturing capability.
+
+## Evidence
+
+- CFS Realta Fusion deal announced April 2026 (TechCrunch) — largest commercial magnet sale to date
+- Type One Energy licensing agreement for stellarator magnets (CFS corporate announcement)
+- University of Wisconsin WHAM magnet supply (CFS/UW partnership)
+- Production rate: 1 pancake/day, >144 of 288 TF pancakes completed for SPARC (CFS Tokamak Times blog)
+- Top 5 REBCO manufacturers control >95% of global HTS tape market (commercial-fusion.beehiiv.com supply chain analysis)
+
+## Challenges
+
+The platform strategy has a tension: selling your best technology to others may erode your competitive advantage. If Realta or Type One achieves fusion with CFS magnets, CFS becomes a supplier rather than the winner. However, this mirrors the NVIDIA playbook — selling picks and shovels during a gold rush is often more profitable than mining. The deeper risk is REBCO tape supply chain concentration: SuperOx (Russian, sanctions-exposed), SuperPower/Furukawa, Fujikura, and AMSC dominate production. A tape shortage constrains everyone, including CFS.
+
+---
+
+Relevant Notes:
+- [[high-temperature superconducting magnets collapse tokamak economics because magnetic confinement scales as B to the fourth power making compact fusion devices viable for the first time]] — the core technology CFS is now selling
+- [[Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue]] — magnet sales bridge the revenue gap
+- [[value in industry transitions accrues to bottleneck positions in the emerging architecture not to pioneers or to the largest incumbents]] — HTS magnets may be the bottleneck position in fusion
+
+Topics:
+- energy systems

domains/energy/CFS magnet pancake production achieved a 30x speedup from 30 days to 1 day per unit suggesting fusion component manufacturing can follow industrial learning curves even if system integration remains unproven.md

@@ -0,0 +1,65 @@
+---
+type: claim
+domain: energy
+description: "CFS achieved 30x production speedup on SPARC magnet pancakes (30 days→1 day), completed >50% of 288 TF pancakes, installed first of 18 magnets January 2026, targeting all 18 by summer 2026 and first plasma 2027"
+confidence: experimental
+source: "Astra, CFS fusion deep dive April 2026; CFS Tokamak Times blog, TechCrunch January 2026, Fortune January 2026"
+created: 2026-04-06
+secondary_domains: ["manufacturing"]
+depends_on:
+  - "Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue"
+  - "high-temperature superconducting magnets collapse tokamak economics because magnetic confinement scales as B to the fourth power making compact fusion devices viable for the first time"
+challenged_by: ["manufacturing speed on identical components does not predict ability to handle integration challenges when 18 magnets, vacuum vessel, cryostat, and plasma heating systems must work together as a precision instrument — ITER's delays happened at integration not component manufacturing"]
+---
+
+# CFS magnet pancake production achieved a 30x speedup from 30 days to 1 day per unit suggesting fusion component manufacturing can follow industrial learning curves even if system integration remains unproven
+
+The dominant narrative about fusion timelines treats the technology as a physics problem — plasma confinement, neutron management, materials science. CFS's SPARC construction data reveals that a significant fraction of the timeline risk is actually a manufacturing problem, and manufacturing problems follow learning curves. However, this evidence is specific to repetitive component production — integration of the complete machine is a fundamentally different challenge.
+
+**The data:**
+- First magnet pancake: 30 days to manufacture
+- 16th pancake: 12 days
+- Current rate: 1 pancake per day
+- Total needed for SPARC: 288 toroidal field pancakes (16 pancakes × 18 D-shaped magnets)
+- Progress: >144 pancakes completed (well over half)
+- Each pancake: steel plate housing REBCO HTS tape in a spiral channel
+- Each assembled magnet: ~24 tons, generating 20 Tesla field
+
+This is a 30x speedup — consistent with manufacturing learning curves observed in automotive, aerospace, and semiconductor fabrication. CFS went through approximately 6 major manufacturing process upgrades to reach this rate. The factory transitioned from artisanal (hand-crafted, one-at-a-time) to industrial (standardized, repeatable, rate-limited by material flow rather than human skill).
+
+**Construction milestones (verified as of January 2026):**
+- Cryostat base installed
+- First vacuum vessel half delivered (48 tons, October 2025)
+- First of 18 HTS magnets installed (January 2026, announced at CES)
+- All 18 magnets targeted by end of summer 2026
+- SPARC nearly complete by end 2026
+- First plasma: 2027
+
+**NVIDIA/Siemens digital twin partnership:** CFS is building a digital twin of SPARC using NVIDIA Omniverse and Siemens Xcelerator, enabling virtual commissioning and plasma optimization. CEO Bob Mumgaard: "CFS will be able to compress years of manual experimentation into weeks of virtual optimization."
+
+This matters for the ARC commercial timeline — but with an important caveat. The pancake production learning curve validates that *component manufacturing* can follow industrial scaling laws. Whether the complete machine assembly, commissioning, and plasma operations also follow such curves is undemonstrated. ITER's decades of delays happened primarily during integration, not during component manufacturing. CFS's compact design (1.85m vs ITER's 6.2m major radius) may simplify integration — or may merely compress the same problems into tighter tolerances.
+
+## Evidence
+
+- 30 days → 12 days → 1 day pancake production rate (CFS Tokamak Times blog, Chief Science Officer Brandon Sorbom)
+- >144 of 288 TF pancakes completed (CFS blog, "well over half")
+- First magnet installed January 2026 (TechCrunch, Fortune, CFS CES announcement)
+- 18 magnets targeted by summer 2026 (Bob Mumgaard, CFS CEO)
+- NVIDIA/Siemens digital twin partnership (CFS press release, NVIDIA announcement)
+- DOE validated magnet performance September 2025, awarding $8M Milestone award
+
+## Challenges
+
+Manufacturing speed on repetitive components (pancakes) is the easiest part of the learning curve. The hardest phases are ahead: integration of 18 magnets into a precision toroidal array, vacuum vessel assembly, cryogenic system commissioning, plasma heating installation, and achieving first plasma. These are one-time engineering challenges that don't benefit from repetitive production learning. ITER's 20-year construction delays happened primarily during integration, not component manufacturing. The true test is whether CFS's compact design genuinely simplifies integration or merely compresses the same problems into tighter tolerances.
+
+The generalization from "pancake production follows learning curves" to "fusion manufacturing follows industrial scaling patterns" is an unsupported leap at this stage. The claim is best understood as evidence that one specific component type at one specific company shows industrial manufacturing characteristics — a necessary but not sufficient condition for the broader thesis.
+
+---
+
+Relevant Notes:
+- [[Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue]] — construction velocity data strengthens timeline credibility
+- [[fusion contributing meaningfully to global electricity is a 2040s event at the earliest because 2026-2030 demonstrations must succeed before capital flows to pilot plants that take another decade to build]] — SPARC is the critical near-term proof point in this timeline
+- [[high-temperature superconducting magnets collapse tokamak economics because magnetic confinement scales as B to the fourth power making compact fusion devices viable for the first time]] — the magnets being manufactured
+
+Topics:
+- energy systems

domains/energy/private fusion has three credible approaches with independent risk profiles where CFS bets on proven tokamak physics Helion on engineering simplicity and TAE on aneutronic fuel.md

@@ -0,0 +1,74 @@
+---
+type: claim
+domain: energy
+description: "CFS (tokamak, HTS magnets, Q~11 target, ARC 400MW early 2030s), Helion (FRC, pulsed non-ignition, direct electricity conversion, Microsoft PPA), and TAE ($1.79B, aneutronic p-B11) represent the three most-capitalized private fusion pathways with fundamentally different risk profiles"
+confidence: experimental
+source: "Astra, CFS fusion deep dive April 2026; CFS corporate, Helion corporate, TAE corporate, FIA 2025 report, TechCrunch, Clean Energy Platform"
+created: 2026-04-06
+secondary_domains: ["space-development"]
+depends_on:
+  - "Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue"
+  - "fusion contributing meaningfully to global electricity is a 2040s event at the earliest because 2026-2030 demonstrations must succeed before capital flows to pilot plants that take another decade to build"
+challenged_by: ["all three could fail for unrelated reasons making fusion portfolio theory moot; Tokamak Energy (UK, spherical tokamak, HTS magnets) and Zap Energy (sheared-flow Z-pinch, no magnets) are also credible contenders; government programs (ITER successor, Chinese CFETR) may solve fusion before any private company"]
+---
+
+# Private fusion has three credible approaches with independent risk profiles where CFS bets on proven tokamak physics Helion on engineering simplicity and TAE on aneutronic fuel
+
+The fusion landscape has 53 companies and $9.77B in cumulative funding (FIA 2025), but three private companies stand out by capitalization and technical credibility: CFS, Helion, and TAE Technologies. They've made fundamentally different technical bets, and understanding the differences is essential for evaluating fusion timelines.
+
+**CFS (Commonwealth Fusion Systems) — the confident physics bet:**
+- **Approach:** Compact tokamak with HTS magnets (proven confinement physics, scaled down via B^4 relationship)
+- **Key advantage:** Tokamak physics is the most studied and best-understood fusion approach. ITER, JET, and decades of government research provide a deep physics basis. CFS's innovation is making tokamaks smaller and cheaper via HTS magnets, not inventing new physics.
+- **Demo:** SPARC at Devens, MA. Q>2 target (models predict Q~11). First plasma 2027.
+- **Commercial:** ARC at James River, Virginia. 400 MW net electrical. Early 2030s. Full output pre-sold (Google + Eni).
+- **Funding:** ~$2.86B raised. Investors include Google, NVIDIA, Tiger Global, Eni, Morgan Stanley.
+- **Risk profile:** Plasma physics risk is LOW (tokamaks are well-understood). Engineering risk is HIGH (tritium breeding, materials under neutron bombardment, thermal conversion, complex plant systems).
+
+**Helion Energy — the engineering simplicity bet:**
+- **Approach:** Field-reversed configuration (FRC) with pulsed, non-ignition plasma. No need for sustained plasma confinement — plasma is compressed, fuses briefly, and the magnetic field is directly converted to electricity.
+- **Key advantage:** No steam turbines. Direct energy conversion (magnetically induced current from expanding plasma) could achieve >95% efficiency. No tritium breeding required if D-He3 fuel works. Dramatically simpler plant design.
+- **Demo:** Polaris (7th prototype) built 2024. Orion (first commercial facility) broke ground July 2025 in Malaga, Washington.
+- **Commercial:** Microsoft PPA. Target: electricity by 2028 (most aggressive timeline in fusion industry).
+- **Funding:** >$1B raised. Backed by Sam Altman (personal, pre-OpenAI CEO), Microsoft, Capricorn Investment Group.
+- **Risk profile:** Engineering risk is LOW (simpler plant, no breeding blankets, direct conversion). Plasma physics risk is HIGH (FRC confinement is less studied than tokamaks, D-He3 fuel requires temperatures 5-10x higher than D-T, limited experimental basis at energy-producing scales).
+
+**TAE Technologies — the aneutronic long shot:**
+- **Approach:** FRC-based, targeting aneutronic proton-Boron-11 (p-B11) fuel — no neutrons means no radioactive activation of reactor walls.
+- **Key advantage:** If it works, no radioactive waste, no tritium supply constraints, no materials degradation from neutron bombardment. Eliminates the hardest engineering problems in fusion.
+- **Demo:** Norman device operational. Copernicus next-gen device planned. Da Vinci commercial target early 2030s.
+- **Funding:** $1.79B raised — second-highest in private fusion after CFS.
+- **Risk profile:** Physics risk is VERY HIGH (p-B11 requires ~3 billion degrees, 20x harder than D-T). Potential reward is correspondingly extreme — truly clean fusion with minimal waste.
+
+**The portfolio insight:** These represent genuinely independent bets. CFS failing (e.g., tritium breeding never scales, materials degrade too fast) does not imply Helion fails (different fuel, different confinement, different conversion). Helion failing (e.g., FRC confinement doesn't scale, D-He3 temperatures unreachable) does not imply TAE fails (different FRC geometry, different fuel target). An investor or policymaker who wants to bet on "fusion" should understand that they're betting on a portfolio of approaches with different failure modes.
+
+**Other credible contenders:**
+- **Tokamak Energy** (UK) — spherical tokamak with HTS magnets, different geometry from CFS, targeting pilot plant mid-2030s
+- **Zap Energy** — sheared-flow Z-pinch, no magnets at all, compact and cheap if physics works
+- **General Fusion** — magnetized target fusion, backed by Jeff Bezos, building demo plant in UK
+
+## Evidence
+
+- CFS: SPARC milestones, $2.86B raised, Google/Eni PPAs, DOE-validated magnets (multiple sources cited in existing CFS claims)
+- Helion: Orion groundbreaking July 2025 in Malaga, WA (Helion press release); Microsoft PPA May 2023; Polaris 7th prototype; Omega manufacturing facility production starting 2026
+- TAE Technologies: $1.79B raised, Norman device operational, UKAEA neutral beam joint venture (TAE corporate, Clean Energy Platform)
+- FIA 2025 industry survey: 53 companies, $9.77B cumulative funding, 4,607 direct employees
+- D-He3 temperature requirements: ~600 million degrees vs ~150 million for D-T (physics constraint)
+- p-B11 temperature requirements: ~3 billion degrees vs ~150 million for D-T (physics constraint)
+
+## Challenges
+
+All three leading companies could fail. Fusion may ultimately be solved by a government program (ITER successor, Chinese CFETR) rather than private companies. The 53 companies and $9.77B represents a venture-capital fusion cycle that could collapse in a funding winter if 2027-2028 demonstrations disappoint — repeating the pattern of earlier fusion hype cycles.
+
+The portfolio framing also obscures a selection effect: private fusion companies have strong incentives to differentiate their pitch to investors, which may exaggerate the independence of their approaches. All face common constraints (plasma physics at scale, materials science, regulatory licensing) that could cause correlated failure across the portfolio.
+
+---
+
+Relevant Notes:
+- [[Commonwealth Fusion Systems is the best-capitalized private fusion company with 2.86B raised and the clearest technical moat from HTS magnets but faces a decade-long gap between SPARC demonstration and commercial revenue]] — the CFS side of this comparison
+- [[high-temperature superconducting magnets collapse tokamak economics because magnetic confinement scales as B to the fourth power making compact fusion devices viable for the first time]] — CFS's core technology advantage
+- [[the gap between scientific breakeven and engineering breakeven is the central deception in fusion hype because wall-plug efficiency turns Q of 1 into net energy loss]] — Helion's direct conversion may avoid this gap entirely
+- [[tritium self-sufficiency is undemonstrated and may constrain fusion fleet expansion because global supply is 25 kg decaying at 5 percent annually while each plant consumes 55 kg per year]] — CFS faces this constraint, Helion's D-He3 and TAE's p-B11 paths avoid it
+- [[fusion contributing meaningfully to global electricity is a 2040s event at the earliest because 2026-2030 demonstrations must succeed before capital flows to pilot plants that take another decade to build]] — all three companies are critical near-term proof points
+
+Topics:
+- energy systems