--- title: Power policy and open weights became the week's leverage while the closed frontier stalled. publication: The AI Stack Weekly slug: 2026-W25 issueNumber: 9 isoYear: 2026 isoWeek: 25 publishedAt: '2026-06-20' canonicalUrl: https://brianletort.ai/industry/weekly/2026-W25 pdfUrl: https://brianletort.ai/downloads/ai-stack-weekly-2026-W25.pdf schemaVersion: 2026.05.02 flywheelArc: all-three capitalFlow: - category: Frontier Labs capitalIn: ~$95B capitalInPrior: ~$95B capitalInDirection: flat revenueOut: ~$21B revenueOutPrior: ~$21B revenueOutDirection: flat burnToRevenue: ~1.3x - category: Hyperscaler-Hosted capitalIn: ~$187B capitalInPrior: ~$185B capitalInDirection: up revenueOut: ~$62B revenueOutPrior: ~$62B revenueOutDirection: flat burnToRevenue: ~0.3x - category: Neoclouds capitalIn: ~$12B capitalInPrior: ~$12B capitalInDirection: flat revenueOut: ~$5B revenueOutPrior: ~$5B revenueOutDirection: flat burnToRevenue: ~3x - category: On-Prem / Hybrid capitalIn: ~$94B capitalInPrior: ~$93B capitalInDirection: up revenueOut: ~$36B revenueOutPrior: ~$36B revenueOutDirection: flat burnToRevenue: ~2x levers: - metric: Frontier lab cash position (avg months runway, top 3) current: ~34-37 mo (flat; no new in-window round) prior: ~34-37 mo (xAI now public via SpaceX IPO) direction: flat threshold: <18 mo triggers re-rating risk - metric: Hyperscaler capex / AI revenue ratio (top 4 weighted) current: ~5.0-5.3 (Amazon $10B + Google $1.5B added; top-4 guides flat) prior: ~5.0-5.3 (Oracle FY27 ~$70B guide; top-4 flat) direction: flat threshold: '>6.0 invites investor pushback at next earnings' - metric: CoreWeave revenue backlog current: ~$100B (Jun 15); Cantor estimate ~$131B by end-Q2 prior: $99.4B (next print Q2, early Aug) direction: up threshold: Conversion velocity matters more than gross figure - metric: NVIDIA Q-over-Q data center revenue current: $75.2B (Q1 FY27); Q2 guide $91B, reports Aug 26 prior: $75.2B (Q1 FY27); Q2 guide $91B, reports Aug 26 direction: flat threshold: Q2 FY27 guide $91B implies further +21% QoQ - metric: Open vs closed gap on coding (SWE-Bench / agentic) current: >- Open closing fast: GLM-5.2 (MIT) reportedly beats GPT-5.5 on long-horizon coding at ~1/6 cost; AA rebased to v4.1 prior: Closed lead widened on paper (Fable 5 80.3% vendor scaffold) vs open ~58.6%; contested direction: down threshold: Sustained open lead reshapes enterprise procurement - metric: Sovereign AI commitments (count / aggregate $) current: ~14 / ~$180B+ (flat; TensorX/Solstice up-to-$1B EU facility is capacity, not drawn) prior: ~14 / ~$180B+ (NAVER+NVIDIA 1GW added) direction: flat threshold: null - metric: PJM 2026/27 capacity auction price ($/MW-day) current: $329.17; 2028/29 BRA results expected ~July 7 (pending) prior: $329.17 (2028/29 BRA clears Jun 30-Jul 7) direction: flat threshold: 11x in 24 months — power is the new binding constraint - metric: Time-to-power, busiest US markets (months) current: 60-84; FERC Jun 18 show-cause may compress large-load study timelines prior: 60-84 (new PJM); 2027/28 auction cleared short of target direction: flat threshold: null - metric: Cost-per-task, frontier reasoning model current: >- ~$0.10-$0.15 (effective); open weights (GLM-5.2) push commodity capability to ~1/6 frontier cost prior: ~$0.10-$0.15 (effective); Fable 5 sets a 2x-Opus top tier ($10/$50 per MTok) direction: down threshold: null - metric: Custom silicon share of incremental AI compute current: ~33-36%; J.P. Morgan pegs 2026 custom-ASIC TAM ~$60-70B (Broadcom 80-85%) prior: ~33-36%; d-Matrix Corsair inference ASIC in volume production direction: up threshold: '>35% materially compresses merchant GPU pricing' predictions: - id: p43-open-weight-top5 lens: software confidencePct: 58 deadline: By August 31, 2026 text: >- An MIT- or Apache-licensed open-weight model (e.g., GLM-5.2) enters the overall top 5 of the Artificial Analysis Intelligence Index v4.1 — not just the open-weight subset — by August 31, 2026. - id: p44-hbm4-allocation-2027 lens: hardware confidencePct: 75 deadline: By August 31, 2026 text: >- By August 31, 2026, all three HBM makers (SK hynix, Samsung, Micron) confirm HBM fully allocated for 2026 and/or 2027 price increases. - id: p45-second-1-6t-design-win lens: networking confidencePct: 55 deadline: By September 30, 2026 text: >- A second non-Broadcom vendor (Marvell or Credo) cites a 1.6T or co-packaged-optics production design win by September 30, 2026. - id: p46-ferc-rto-compliance lens: power confidencePct: 80 deadline: By August 17, 2026 text: >- At least one RTO/ISO files a large-load interconnection compliance proposal answering FERC's Jun 18 show-cause orders by the August 17, 2026 deadline. - id: p47-hyperscaler-1gw-btm lens: power confidencePct: 68 deadline: By August 31, 2026 text: >- A named hyperscaler announces a >1GW behind-the-meter or off-grid generation deal for AI data centers by August 31, 2026. - id: p48-closed-price-response lens: capital confidencePct: 55 deadline: By August 31, 2026 text: >- At least one major closed lab cuts flagship API prices or ships a cheaper tier by August 31, 2026, in response to open-weight cost pressure. predictionsPrior: - id: p37-gemini-3-5-pro-below-fable5 lens: software outcome: pending deadline: By June 30, 2026 text: >- Gemini 3.5 Pro reaches public GA by June 30, 2026 but debuts below Claude Fable 5 on the Artificial Analysis Intelligence Index in its first independent pass. - id: p38-fable5-access-restored lens: software outcome: pending deadline: By July 31, 2026 text: >- Anthropic restores public (non-government) access to a Fable-5-class model — with geo-gating or KYC controls — by July 31, 2026. - id: p39-second-inference-asic-milestone lens: hardware outcome: pending deadline: By August 31, 2026 text: >- At least one more merchant inference-ASIC vendor (Groq, Cerebras, Tenstorrent, or d-Matrix) announces a volume-production or named-hyperscaler milestone before August 31, 2026. - id: p40-second-1-6t-design-win lens: networking outcome: pending deadline: By August 31, 2026 text: >- A second of (Marvell, Credo) cites a 1.6T or co-packaged-optics production design win by August 31, 2026, joining Broadcom's Tomahawk 6. - id: p41-no-public-s1 lens: capital outcome: pending deadline: By August 31, 2026 text: >- Neither OpenAI nor Anthropic converts its confidential draft to a publicly visible S-1 on SEC EDGAR before August 31, 2026. - id: p42-hyperscaler-power-first lens: power outcome: pending deadline: By September 30, 2026 text: >- A hyperscaler announces a >500MW power-first AI campus or behind-the-meter generation deal by September 30, 2026. signalScores: - 5 - 4 - 4 - 2 - 2 keyTakeaways: - >- FERC issued six simultaneous Section 206 show-cause orders telling PJM, MISO, SPP, CAISO, ISO-NE, and NYISO their tariffs 'appear unjust and unreasonable' for large data-center loads — national interconnection rules are being rewritten on a 60-day clock. - >- Z.ai's GLM-5.2 shipped under a genuine MIT license, and independent testing says it beats GPT-5.5 on several long-horizon coding benchmarks at roughly one-sixth the cost — while Claude Fable 5 stayed government-suspended the entire week and Gemini 3.5 Pro slipped to July. - >- Developers kept routing around the interconnection queue entirely (Cummins-Circe's 2GW behind-the-meter gas) while Amazon ($10B Missouri) and Google ($1.5B Alabama) added grid-connected capacity explicitly paying 100% of interconnection cost. - >- Capital flow: CoreWeave's backlog ticked to ~$100B with a Cantor estimate of ~$131B by end-Q2; Meta-Crusoe's reported ~1.6GW deal stayed unconfirmed with both parties declining comment. - >- Signal-vs-noise: GPT-5.6 and 'Grok V9' imminence is rumor with no primary artifact; treat the Meta-Crusoe deal as reported, not closed. - >- Watch PJM's 2028/29 capacity auction results (~Jul 7) and the RTO/ISO compliance filings due by Aug 17 — the quarter's most important power-cost signals. byTheNumbers: - value: '6' label: Simultaneous FERC Section 206 show-cause orders - value: ~1/6 label: GLM-5.2's cost versus comparable frontier models - value: 2GW label: Cummins behind-the-meter gas for Circe's West Texas campuses - value: $10B label: Amazon's Montgomery County, MO data-center campus - value: ~$100B label: CoreWeave backlog as of mid-June - value: $60-70B label: J.P. Morgan's 2026 custom-ASIC market estimate --- # Power policy and open weights became the week's leverage while the closed frontier stalled. *Issue 09 · Week 25 of 2026 · Published 2026-06-20* ## Executive summary - FERC issued six simultaneous Section 206 show-cause orders telling PJM, MISO, SPP, CAISO, ISO-NE, and NYISO their tariffs 'appear unjust and unreasonable' for large data-center loads — national interconnection rules are being rewritten on a 60-day clock. - Z.ai's GLM-5.2 shipped under a genuine MIT license, and independent testing says it beats GPT-5.5 on several long-horizon coding benchmarks at roughly one-sixth the cost — while Claude Fable 5 stayed government-suspended the entire week and Gemini 3.5 Pro slipped to July. - Developers kept routing around the interconnection queue entirely (Cummins-Circe's 2GW behind-the-meter gas) while Amazon ($10B Missouri) and Google ($1.5B Alabama) added grid-connected capacity explicitly paying 100% of interconnection cost. - Capital flow: CoreWeave's backlog ticked to ~$100B with a Cantor estimate of ~$131B by end-Q2; Meta-Crusoe's reported ~1.6GW deal stayed unconfirmed with both parties declining comment. - Signal-vs-noise: GPT-5.6 and 'Grok V9' imminence is rumor with no primary artifact; treat the Meta-Crusoe deal as reported, not closed. - Watch PJM's 2028/29 capacity auction results (~Jul 7) and the RTO/ISO compliance filings due by Aug 17 — the quarter's most important power-cost signals. **By the numbers.** - **6** — Simultaneous FERC Section 206 show-cause orders (Every major US RTO/ISO must answer within 60 days) - **~1/6** — GLM-5.2's cost versus comparable frontier models (MIT-licensed; reportedly beats GPT-5.5 on long-horizon coding) - **2GW** — Cummins behind-the-meter gas for Circe's West Texas campuses (Bypassing the ERCOT interconnection queue entirely) - **$10B** — Amazon's Montgomery County, MO data-center campus (Paying 100% of interconnection cost amid FERC cost-shift scrutiny) - **~$100B** — CoreWeave backlog as of mid-June (Cantor Fitzgerald models ~$131B by end-Q2 — analyst-estimated) - **$60-70B** — J.P. Morgan's 2026 custom-ASIC market estimate (Broadcom at 80-85% share, Marvell 10-12%) ## Big Story W25 moved at the two ends of the stack that the AI factory actually depends on — the grid and the open model layer — while the closed frontier marked time. The single most consequential event was regulatory: on Jun 18 FERC issued six simultaneous Section 206 show-cause orders telling PJM, MISO, SPP, CAISO, ISO-NE, and NYISO their tariffs 'appear unjust and unreasonable' for large data-center loads, forcing co-location/behind-the-meter rules and cost-shift transparency within 60 days. Developers kept routing around the interconnection queue entirely (Cummins–Circe's 2GW behind-the-meter gas) while hyperscalers added grid-connected capacity explicitly paying 100% of interconnection cost (Amazon's $10B Missouri campus, Google's $1.5B Alabama expansion). At the model layer the story was openness and stasis: Z.ai shipped GLM-5.2 under a genuine MIT license, which independent testing says beats GPT-5.5 on several long-horizon coding benchmarks at roughly one-sixth the cost; MiniMax-M3's sparse-attention weights matured; and Artificial Analysis rebased its Intelligence Index to v4.1 around agentic tasks — all while Anthropic's Claude Fable 5 stayed government-suspended the entire week (Opus 4.8 remained the working leader) and Gemini 3.5 Pro slipped to July. Hardware and networking were quiet (Supermicro opened Vera Rubin NVL72 order availability; Coherent expanded indium-phosphide laser capacity). Net/net: boards should treat power policy as a board-level siting and cost risk now being repriced by FERC; investors should watch the open-weight cost curve collapsing frontier-adjacent capability toward commodity inference; architects should pilot MIT-licensed GLM-5.2 for sovereign and self-host coding while keeping multi-model fallbacks given the closed frontier's demonstrated fragility; operators should lock generation and long-lead transformers before GPUs. Flywheel arc: `all-three`. ## Software lens - **Jun 16.** Z.ai released GLM-5.2 under an MIT license (~744B / ~40B-active sparse-attention MoE, 1M context); independent testing reports it beats GPT-5.5 on several long-horizon coding benchmarks at ~1/6 the cost, and Artificial Analysis cites it as the leading open-weight model _([Z.ai blog; VentureBeat; Hugging Face](https://z.ai/blog/glm-5.2))_ - **Jun 15.** Artificial Analysis rebased its Intelligence Index to v4.1, re-weighting the industry's headline benchmark around agentic tasks (GDPval-AA v2, Terminal-Bench, banking agents); scores are not back-comparable to v4.0 _([Artificial Analysis](https://artificialanalysis.ai/articles/artificial-analysis-intelligence-index-v4-1))_ - **Jun 15-21.** Anthropic's Claude Fable 5 and Mythos 5 remained government-suspended for the full week with no restoration; Opus 4.8 stayed the working closed-frontier leader as queries continued to fall back to it _([Anthropic](https://www.anthropic.com/news/claude-fable-5-mythos-5))_ - **Jun 18.** Gemini 3.5 Pro slipped its GA from June to July: still a limited Vertex preview with no public model card, pricing, or independent benchmark vs Fable 5/Opus 4.8 _([Business Insider; Google](https://www.businessinsider.com/google-3-5-pro-july-release-tokens-ai-agents-model-2026-6))_ **What this means.** The closed frontier paused — no GA from OpenAI, Google, or a restored Anthropic — and the momentum shifted to open weights and to how 'best' is measured. Architects should pilot MIT-licensed GLM-5.2 for self-host/sovereign coding now and re-baseline on the agentic AA v4.1 index; see The Model Pulse for the full architecture read. ## Hardware lens - **Jun 17.** Supermicro opened customer engagement / order availability for Vera Rubin NVL72 and HGX Rubin NVL8 reference designs (1,152-GPU scalable unit, 331TB HBM4), though deployments remain H2 2026 at GA — order availability, not shipments _([The Data Center Engineer (Supermicro DCBBS)](https://thedatacenterengineer.com/news/supermicro-dcbbs-blueprints-scale-nvidia-vera-rubin-ai-factories-from-5mw-to-1gw/))_ - **Jun 16.** HPE Discover expanded Private Cloud AI with NVIDIA Vera-CPU servers (ProLiant DL394) and the NVIDIA Agent Toolkit — on-prem agentic-AI enablement rather than new silicon or shipments _([SiliconANGLE](https://siliconangle.com/2026/06/16/hpe-expands-private-cloud-ai-factory-portfolio-support-next-gen-autonomous-agents/))_ - **Jun 18.** A J.P. Morgan note pegged the 2026 custom-ASIC market at ~$60-70B with Broadcom at 80-85% and Marvell 10-12% share, reinforcing the ASIC-overtakes-merchant-GPU-units thesis for 2027 _([J.P. Morgan note (Finvaulta summary)](https://finvaulta.com/research/jp-morgan/ai-drives-resurgence-in-custom-chips-2026-06-18))_ **What this means.** Hardware was a follow-through week: the marquee catalysts (Micron earnings, SK hynix HBM4 dynamics, NVIDIA ISC, Groq's raise) all landed Jun 22-24, just outside the window. In-window the signal was OEM/system enablement (Supermicro order availability, HPE Vera CPU) and the custom-silicon thesis hardening. Operators should treat Vera Rubin as orderable but not shippable until H2; investors should keep diversifying toward the Broadcom/Marvell ASIC duopoly. ## Networking lens - **Jun 16.** Coherent broke ground on a Sherman, TX indium-phosphide fab expansion and signed a $50M CHIPS LOI to roughly quadruple InP wafer output — the lasers feeding 800G-to-1.6T transceivers and NVIDIA CPO external laser sources _([Coherent; NVIDIA blog; The Register](https://www.coherent.com/news/press-releases/a-chip-letter-of-intent-for-50m-to-expand-world-leading-manufacturing-facility-for-ai-infrastructure))_ - **Jun 17.** HPE Juniper began shipping the QFX5250-64OE-L, an industry-first 1.6T, fully liquid-cooled 102.4T switch (64x 1.6T ports, ORV3-compliant) built on Broadcom-class silicon _([Juniper community blog](https://community.juniper.net/blogs/riti-sharma/2026/06/17/industry-first-16t-liquid-cooled-sw-qfx5250-64oe-l))_ - **Jun 17.** Marvell published a plasmonics optical-roadmap (with Polariton/ETH Zurich) citing a 1.1THz modulator lab record and 400G/lane SiPho devices — a forward R&D signal, not a commercial design win _([Marvell blog](https://www.marvell.com/blogs/plasmonics-higher-bandwidth-optics-ai-era.html))_ **What this means.** Networking was quiet between the March (OFC) and June-1 (earnings) news clusters: the week's signal was supply-chain capacity (Coherent's InP expansion addressing the laser chokepoint) plus a liquid-cooled 1.6T system, not a new chip design win. Critically, no second non-Broadcom vendor (Marvell or Credo) cited a fresh 1.6T/CPO production win — architects pressing for a second silicon source still have to wait. ## Capital flow | Category | Capital in | Revenue out | Burn:Revenue | Movement | |---|---|---|---|---| | Frontier Labs (OpenAI, Anthropic, Google DeepMind, xAI) | ~$95B (was ~$95B, flat) | ~$21B (was ~$21B, flat) | ~1.3x | No new frontier-lab round closed; the action moved to open weights (GLM-5.2 MIT) and to Fable 5 staying government-suspended, not to lab balance sheets. | | Hyperscaler-Hosted (Azure-OpenAI, AWS-Anthropic, Google Cloud-Gemini, Oracle-OCI) | ~$187B (was ~$185B, up) | ~$62B (was ~$62B, flat) | ~0.3x | Amazon ($10B Missouri) and Google ($1.5B Alabama) added grid-connected capacity, both explicitly paying 100% of interconnection cost amid FERC cost-shift scrutiny. | | Neoclouds (CoreWeave, Nscale, Crusoe, Lambda, Fluidstack, IREN) | ~$12B (was ~$12B, flat) | ~$5B (was ~$5B, flat) | ~3x | CoreWeave's backlog ticked to ~$100B with a Cantor estimate of ~$131B by end-Q2; Meta–Crusoe's reported ~1.6GW deal stayed unconfirmed. | | On-Prem / Hybrid (Enterprise GPU clusters, sovereign and national programs, Cisco / Dell / HPE) | ~$94B (was ~$93B, up) | ~$36B (was ~$36B, flat) | ~2x | FERC's six show-cause orders put national interconnection rules in play; Cummins–Circe committed 2GW of behind-the-meter gas, and TensorX/Solstice lined up up-to-$1B of EU sovereign-AI financing. | ### Frontier Labs — detail Frontier-lab capital was quiet in-window. No new mega-round priced, and both OpenAI (Jun 8) and Anthropic (Jun 1) remain at the confidential-draft-S-1 stage with nothing publicly visible on EDGAR. The competitive pressure this week came from the open-weight side — GLM-5.2's MIT release at ~1/6 the cost of comparable frontier models — rather than from new financing. Treat this row as capital already committed; the next regime change is a public S-1. **Transactions:** - **Jun 15-21.** No new frontier-lab round; OpenAI and Anthropic S-1s remain confidential drafts — No public S-1 _([CNBC; OpenAI; Anthropic](https://www.cnbc.com/2026/06/08/openai-confidentially-files-for-ipo-prepping-wall-street-for-ai-debut.html))_ ### Hyperscaler-Hosted — detail Hyperscaler build-out continued with two incremental campus commitments — Amazon's $10B Montgomery County, MO site and Google's $1.5B Jackson County, AL expansion — both framed around paying full interconnection cost so ratepayers are not burdened. That framing is now standard messaging under FERC's cost-shift scrutiny. The aggregate ~$725-805B 2026 capex guide is carried flat (no new top-4 print in-window); these are demand-led capacity adds, not a strategy change. **Transactions:** - **Jun 15.** Amazon $10B data-center campus, Montgomery County, MO (pays 100% interconnection) — $10B _([Missouri DED](https://ded.mo.gov/press-room/amazon-investing-10-billion-montgomery-county))_ - **Jun 16.** Google $1.5B Jackson County, AL expansion (>300MW contracted, Kairos SMR partner) — $1.5B _([Alabama Reporter](https://www.alreporter.com/2026/06/16/google-plans-1-5-billion-jackson-county-data-center-expansion/))_ ### Neoclouds — detail The neocloud row held flat on fresh capital but the backlog signal firmed: company-adjacent reporting puts CoreWeave's backlog at ~$100B as of mid-June (vs the $99.4B Q1 figure), with Cantor Fitzgerald modeling ~$131B by end-Q2 — analyst-estimated, not company-guided. Meta's reported ~1.6GW Crusoe deal (Childress TX + Warrenton MO) remained unconfirmed with both parties declining comment. Operators should keep weighting conversion velocity over headline backlog. **Transactions:** - **Jun 15.** CoreWeave backlog ~$100B (Cantor estimate ~$131B by end-Q2) — ~$100B (est. ~$131B) _([Inc. / Barron's (Cantor estimate)](https://www.inc.com/georgia-fearn/core-weave-is-quietly-stacking-131-billion-backlog-behind-the-scenes/91361642))_ - **Jun 18.** Meta–Crusoe ~1.6GW (reported, unconfirmed; both declined comment) — ~1.6GW (reported) _([Bloomberg](https://www.bloomberg.com/news/articles/2026-06-18/meta-strikes-new-ai-computing-deals-with-data-centers))_ ### On-Prem / Hybrid — detail The on-prem/sovereign row is increasingly a power-and-permitting story. FERC's Jun 18 orders force every major RTO/ISO to rewrite large-load interconnection within 60 days, while developers keep building their own generation — Cummins will supply 2GW of behind-the-meter gas to Circe's West Texas campuses, bypassing the ERCOT queue. In Europe, TensorX and Solstice lined up an up-to-$1B GPU/data-center financing facility (capacity, not a drawn commitment). Sovereign and on-prem buyers should assemble firm power and permitting before announcing GPU counts. **Transactions:** - **Jun 18.** FERC issues six Section 206 show-cause orders rewriting large-load interconnection — All US RTOs/ISOs (60-day clock) _([FERC; Day Pitney](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration))_ - **Jun 16.** Cummins to supply 2GW behind-the-meter gas to Circe Energy (West Texas) — 2GW BTM _([mgrid](https://mgrid.org/2026/06/16/cummins-to-power-circe-energys-west-texas-ai-data-centers-with-2-gw-of-behind-the-meter-gas-generation/))_ ## Signal vs noise - **Score 5/5 —** FERC issued six simultaneous Section 206 show-cause orders rewriting large-load/data-center interconnection across all US RTOs/ISOs. - _Sources:_ FERC press release and orders; Day Pitney and Akin/JDSupra legal analyses. Primary regulatory action, multi-source. - _Read:_ SIGNAL. The rules for how AI campuses connect and who pays are being rewritten nationally on a 60-day clock. This is a board-level siting and cost variable — pro-speed medium-term, but it injects near-term regulatory uncertainty into every US data-center plan. - **Score 4/5 —** Z.ai's GLM-5.2 shipped MIT open weights that beat GPT-5.5 on several long-horizon coding benchmarks at ~1/6 the cost and lead the open-weight field. - _Sources:_ Z.ai release and Hugging Face card (primary); VentureBeat independent testing; Artificial Analysis placement. Caveat: SWE-Bench/Aider numbers not yet fully settled. - _Read:_ SIGNAL. A genuinely permissive (MIT, no regional limits) frontier-adjacent open model is now self-hostable at a fraction of closed-API cost. Architects should pilot it for sovereign/self-host coding; it pressures closed-flagship pricing. - **Score 4/5 —** Anthropic's Claude Fable 5 and Mythos 5 remained globally suspended for the entire week following the Jun 12 government order. - _Sources:_ Anthropic news page (primary) plus secondary trackers; no restoration date announced. - _Read:_ SIGNAL. The sovereign/regulatory single-point-of-failure on a top-tier closed model is not a one-day outage — it persisted a full week. Keep Opus 4.8/Sonnet fallbacks wired and do not single-source the frontier for production-critical paths. - **Score 2/5 —** GPT-5.6 (OpenAI) and Grok V9 (xAI) are launching imminently and will reset the frontier. - _Sources:_ Backend codenames, transient routing strings, and prediction-market odds only; no system card or API contract. GPT-5.5 remains the documented flagship; 'Grok V9' does not exist. - _Read:_ NOISE. Rumor with no primary artifact. Keep current baselines and do not delay procurement on an unreleased SKU; any GPT-5.6 was tracking to land after this window. - **Score 2/5 —** Meta has signed a ~1.6GW compute deal with Crusoe. - _Sources:_ Bloomberg-sourced reporting (Jun 18); both Meta and Crusoe declined to comment, terms undisclosed. - _Read:_ NOISE / unconfirmed. Treat as reported, not closed. The structural pattern (Meta adding external behind-the-meter compute) is real, but the specific deal is not yet verifiable for a procurement or investment decision. ## Levers | Metric | Current | Prior | Direction | Threshold | |---|---|---|---|---| | Frontier lab cash position (avg months runway, top 3) | ~34-37 mo (flat; no new in-window round) | ~34-37 mo (xAI now public via SpaceX IPO) | flat | <18 mo triggers re-rating risk | | Hyperscaler capex / AI revenue ratio (top 4 weighted) | ~5.0-5.3 (Amazon $10B + Google $1.5B added; top-4 guides flat) | ~5.0-5.3 (Oracle FY27 ~$70B guide; top-4 flat) | flat | >6.0 invites investor pushback at next earnings | | CoreWeave revenue backlog | ~$100B (Jun 15); Cantor estimate ~$131B by end-Q2 | $99.4B (next print Q2, early Aug) | up | Conversion velocity matters more than gross figure | | NVIDIA Q-over-Q data center revenue | $75.2B (Q1 FY27); Q2 guide $91B, reports Aug 26 | $75.2B (Q1 FY27); Q2 guide $91B, reports Aug 26 | flat | Q2 FY27 guide $91B implies further +21% QoQ | | Open vs closed gap on coding (SWE-Bench / agentic) | Open closing fast: GLM-5.2 (MIT) reportedly beats GPT-5.5 on long-horizon coding at ~1/6 cost; AA rebased to v4.1 | Closed lead widened on paper (Fable 5 80.3% vendor scaffold) vs open ~58.6%; contested | down | Sustained open lead reshapes enterprise procurement | | Sovereign AI commitments (count / aggregate $) | ~14 / ~$180B+ (flat; TensorX/Solstice up-to-$1B EU facility is capacity, not drawn) | ~14 / ~$180B+ (NAVER+NVIDIA 1GW added) | flat | — | | PJM 2026/27 capacity auction price ($/MW-day) | $329.17; 2028/29 BRA results expected ~July 7 (pending) | $329.17 (2028/29 BRA clears Jun 30-Jul 7) | flat | 11x in 24 months — power is the new binding constraint | | Time-to-power, busiest US markets (months) | 60-84; FERC Jun 18 show-cause may compress large-load study timelines | 60-84 (new PJM); 2027/28 auction cleared short of target | flat | — | | Cost-per-task, frontier reasoning model | ~$0.10-$0.15 (effective); open weights (GLM-5.2) push commodity capability to ~1/6 frontier cost | ~$0.10-$0.15 (effective); Fable 5 sets a 2x-Opus top tier ($10/$50 per MTok) | down | — | | Custom silicon share of incremental AI compute | ~33-36%; J.P. Morgan pegs 2026 custom-ASIC TAM ~$60-70B (Broadcom 80-85%) | ~33-36%; d-Matrix Corsair inference ASIC in volume production | up | >35% materially compresses merchant GPU pricing | **Lever detail:** - **Frontier lab cash position (avg months runway, top 3).** Top 3 frontier labs (OpenAI, Anthropic, Google DeepMind) by disclosed runway, with xAI now public inside SpaceX. No new financing event in-window; both OpenAI and Anthropic remain at the confidential-draft-S-1 stage. Capital access stays wide; boards should not assume funding pressure forces near-term commercial concessions. - **Hyperscaler capex / AI revenue ratio (top 4 weighted).** Top 4 hyperscalers (MSFT, GOOG, META, AMZN) weighted aggregate of capex divided by AI-attributable revenue. No new top-4 earnings print in-window; the incremental Amazon and Google campus commitments fit prior ~$725-805B 2026 capex guides. Investors should keep the bubble hypothesis on funding/conversion and now on FERC-driven siting cost, not demand. - **CoreWeave revenue backlog.** Booked but unrecognized revenue. Reporting puts backlog at ~$100B as of mid-June (vs the $99.4B Q1 figure), with Cantor Fitzgerald modeling ~$131B by end-Q2 — analyst-estimated, not company-guided. The official next print is Q2 in early August. Operators should keep watching conversion velocity over the headline figure. - **NVIDIA Q-over-Q data center revenue.** No within-window NVIDIA event; the next earnings print is Aug 26. The Q2 guide is $91B. In-window context (Supermicro Vera Rubin order availability, all three HBM makers volume-shipping HBM4 12-Hi) supports the ramp; packaging and power remain the binding constraints, not demand. - **Open vs closed gap on coding (SWE-Bench / agentic).** The gap narrowed sharply in-window: GLM-5.2's MIT weights reportedly beat GPT-5.5 on several long-horizon coding benchmarks at ~1/6 the cost, and Artificial Analysis rebased its Intelligence Index to v4.1 (agentic), where open leaders sit ~44 and GLM-5.2 took the open lead. With Fable 5 suspended, the available closed leader is Opus 4.8 (AA v4.1 56). Architects should treat open self-host as a live procurement option, not a hedge. - **Sovereign AI commitments (count / aggregate $).** Analyst-curated count of sovereign/national AI-compute commitments. No major new drawn commitment in-window; the only new item is TensorX/Solstice's up-to-$1B EU GPU/data-center financing facility (capacity, not a drawn commitment). Operators with EU workloads should track the facility but price in multi-year build timelines. - **PJM 2026/27 capacity auction price ($/MW-day).** The 2026/27 BRA cleared at the FERC cap ($329.17); the 2027/28 BRA cleared at $333.44. The 2028/29 auction is still pending, with results expected around July 7, 2026, under a collar floor of $175 and a cap near $325. Architects should not assume near-term price relief; budget capacity at-cap through 2028. - **Time-to-power, busiest US markets (months).** Months from new-load interconnection request to energization. FERC's Jun 18 show-cause orders aim to speed large-load studies and clarify cost allocation within 60 days — potentially pro-speed medium-term, but no near-term change. Large-power-transformer lead times remain ~128 weeks. Architects should pre-commit power and long-lead grid equipment before GPU SKUs. - **Cost-per-task, frontier reasoning model.** Median cost across frontier-tier reasoning models for a benchmark complex task. Open weights drove the ceiling down: GLM-5.2 lists at ~$1.40/$4.40 per MTok (~1/6 of comparable frontier) and Grok 4.3 went GA on Bedrock at $1.25/$2.50. Operators running agents at scale should re-benchmark on cost-per-task and pilot open self-host for routine work. - **Custom silicon share of incremental AI compute.** A J.P. Morgan note put the 2026 custom-ASIC market at ~$60-70B with Broadcom at 80-85% and Marvell 10-12%, reinforcing the thesis that ASIC units overtake merchant-GPU units by 2027. Investors with concentrated NVIDIA exposure should keep diversifying into the Broadcom/Marvell co-design duopoly and advanced packaging / power. ## Predictions - **`p43-open-weight-top5` _[software]_ — An MIT- or Apache-licensed open-weight model (e.g., GLM-5.2) enters the overall top 5 of the Artificial Analysis Intelligence Index v4.1 — not just the open-weight subset — by August 31, 2026.** - Confidence: 58%. Deadline: By August 31, 2026. - Trigger: Artificial Analysis Intelligence Index v4.1 leaderboard update. - **`p44-hbm4-allocation-2027` _[hardware]_ — By August 31, 2026, all three HBM makers (SK hynix, Samsung, Micron) confirm HBM fully allocated for 2026 and/or 2027 price increases.** - Confidence: 75%. Deadline: By August 31, 2026. - Trigger: Earnings calls or supply-chain reporting (TrendForce, Bloomberg/Reuters) from the three memory vendors. - **`p45-second-1-6t-design-win` _[networking]_ — A second non-Broadcom vendor (Marvell or Credo) cites a 1.6T or co-packaged-optics production design win by September 30, 2026.** - Confidence: 55%. Deadline: By September 30, 2026. - Trigger: Earnings call, product release, or customer design-win disclosure. - **`p46-ferc-rto-compliance` _[power]_ — At least one RTO/ISO files a large-load interconnection compliance proposal answering FERC's Jun 18 show-cause orders by the August 17, 2026 deadline.** - Confidence: 80%. Deadline: By August 17, 2026. - Trigger: FERC docket filings from PJM, MISO, SPP, CAISO, ISO-NE, or NYISO. - **`p47-hyperscaler-1gw-btm` _[power]_ — A named hyperscaler announces a >1GW behind-the-meter or off-grid generation deal for AI data centers by August 31, 2026.** - Confidence: 68%. Deadline: By August 31, 2026. - Trigger: Hyperscaler energy/data-center announcement; utility or developer disclosure. - **`p48-closed-price-response` _[capital]_ — At least one major closed lab cuts flagship API prices or ships a cheaper tier by August 31, 2026, in response to open-weight cost pressure.** - Confidence: 55%. Deadline: By August 31, 2026. - Trigger: Lab pricing page or API changelog (OpenAI, Anthropic, Google). ### Prior predictions scored - `p37-gemini-3-5-pro-below-fable5` _[software]_ — **PENDING** — Gemini 3.5 Pro reaches public GA by June 30, 2026 but debuts below Claude Fable 5 on the Artificial Analysis Intelligence Index in its first independent pass. — Tracking toward a miss on the GA leg: as of W25 close, Pro slipped from June to July, still a limited Vertex preview with no public benchmark. Deadline June 30 not yet passed but unlikely to be met. - `p38-fable5-access-restored` _[software]_ — **PENDING** — Anthropic restores public (non-government) access to a Fable-5-class model — with geo-gating or KYC controls — by July 31, 2026. — No restoration in-window — Fable 5 and Mythos 5 stayed government-suspended the full week, with Anthropic saying only that it is 'working to restore access.' Deadline July 31 open. - `p39-second-inference-asic-milestone` _[hardware]_ — **PENDING** — At least one more merchant inference-ASIC vendor (Groq, Cerebras, Tenstorrent, or d-Matrix) announces a volume-production or named-hyperscaler milestone before August 31, 2026. — In-window the only item was a Qualcomm-Tenstorrent acquisition rumor (Jun 15). Strong out-of-window evidence followed Jun 22-24 (Cerebras' AWS/OpenAI confirmations, Groq's $650M raise), so this is tracking to hit. Deadline Aug 31 open. - `p40-second-1-6t-design-win` _[networking]_ — **PENDING** — A second of (Marvell, Credo) cites a 1.6T or co-packaged-optics production design win by August 31, 2026, joining Broadcom's Tomahawk 6. — No new non-Broadcom design win in-window; Marvell's only output was a plasmonics research blog and Coherent's move was laser capacity, not a design win. Deadline Aug 31 open. - `p41-no-public-s1` _[capital]_ — **PENDING** — Neither OpenAI nor Anthropic converts its confidential draft to a publicly visible S-1 on SEC EDGAR before August 31, 2026. — Holding and on track: both remain confidential drafts with no public EDGAR S-1 in-window; a public filing is expected only days before a roadshow (late Q3/Q4). Deadline Aug 31 open. - `p42-hyperscaler-power-first` _[power]_ — **PENDING** — A hyperscaler announces a >500MW power-first AI campus or behind-the-meter generation deal by September 30, 2026. — Strong momentum: in-window power-first activity was developer-led (Cummins-Circe 2GW BTM) with a reported-but-unconfirmed Meta-Crusoe ~1.6GW deal. A qualifying hyperscaler deal (Microsoft-Chevron Project Kilby, 2.67GW) landed Jun 22, just after this window. Deadline Sep 30 open; tracking to hit. ## Watchlist - **Jun 22-Jul 7 — PJM 2028/29 base residual capacity auction results.** After two near-cap clears and a 2027/28 reliability shortfall, the 2028/29 results (collar near $325/MW-day) are the single most important power-cost signal of the quarter and will reprice US data-center economics. - **By Aug 17 — RTO/ISO responses to FERC's Section 206 show-cause orders.** FERC gave six grid operators 60 days to fix large-load interconnection rules. The compliance filings will define how AI campuses connect and who pays — a board-level siting variable for every US deployment. - **Jun 22-24 (W26 catalysts) — Project Kilby, Micron/SK hynix HBM4, Cerebras/Groq.** A cluster of marquee items landed just after this window: Microsoft-Chevron's 2.67GW off-grid gas deal, Micron's record HBM-allocated quarter, SK hynix's HBM4-vs-DDR5 pivot, and Cerebras/Groq milestones. They will anchor next week's read. - **July — Gemini 3.5 Pro GA and Fable 5 restoration.** Both slipped: Pro moved to July and Fable 5 stayed suspended. Their resolution will decide whether the closed frontier re-takes the lead from the surging open-weight field. - **Jun-Aug — Open-weight enterprise adoption (GLM-5.2).** Downloads, integrations, and independent SWE-Bench replication for GLM-5.2 and MiniMax-M3 will show whether MIT/permissive open weights become production substrate and force closed-flagship price cuts. ## Changelog - Reframed W25 around power policy (FERC's six Section 206 show-cause orders) and the open-weight surge (GLM-5.2 MIT) as the week's leverage points while the closed frontier stalled (Fable 5 suspended, Gemini slipped to July). --- Source of truth: `src/data/industry/weekly/2026-W25.ts`. Canonical HTML: . PDF: .