The update loop’s working notes

What sets the pace of AI data centre buildout in the United States. These notes move nothing: only a sealed change, made by the rule, moves a number.

Developments first reported since the record began

These are the update loop's working notes: each development it read, how it read it, and what the published rule makes of that reading. They are not part of the sealed record and they move nothing. Only a sealed change, made by the rule, moves a number.

Since the record began, the update loop has read 6 developments first reported after it began, in 7 readings, one for each assessment a development bears on. 7 readings are waiting for a second outlet, and 0 were set aside.

The newest daily state that records a reading, of 14 October 2026, counts 7 developments waiting for a second outlet. These notes count the same. Both count a development once for each assessment it bears on. The daily states also count reports, while these notes list developments, and a report retrieved after the record began but first reported before it is counted in the daily states and never weighed by the rule.

Each outcome is worked out by the published rule from what the loop recorded when it read the development: its outlets then, its reading, and the challenge step's verdict once that has run. The outlet grades are the loop's own.

• Read 14 October 2026, 09:46 UTC · Main assessment
  • “Utilities Struggle to Confront Issue of ‘Phantom' AI Data Centers” · breitbart.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M1 (Waits for large grid connections). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report states utilities are tightening connection requirements as speculative data-center requests swell grid queues, directly indicating that grid interconnection queues are a real and growing bottleneck for data centre buildout, strengthening M1. This does not support the chip-supply or capital alternatives, and the queue issue is distinct from local permitting.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 14 October 2026, 09:46 UTC · The condition it rests on
  • “Hyperscale Data Reduces ATM Program, Shifts to Debt” · tipranks.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R3 (Financing for data centre campuses). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: more likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Hyperscale Data is closing project debt financing for a Michigan AI data-centre campus serving a neocloud customer, which is a concrete financing event keeping construction moving. This fits R3 (debt financing closing on schedule) and also supports RX2, as the capacity is being built outside the largest cloud companies' balance sheets through a specialized provider financed via a special-purpose debt structure.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 14 October 2026, 09:46 UTC · Main assessment
  • “FERC approves Duke large load transmission rules despite cost-shift concerns” · utilitydive.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: medium.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC formally approved Duke Energy Carolinas' tariff provisions imposing financial security requirements and special terms on large new loads, which is a regulatory order rationing connection capacity and directly instantiates M2. This strengthens the case that grid access and interconnection terms are a binding constraint, with no indication that chips, capital, or local permitting are the operative issue here.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 14 October 2026, 09:46 UTC · Main assessment
  • “Federal energy regulators shoot down grid operator's plan to cut power to data centers during shortfalls” · thehill.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: medium.
  Weakens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: less likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC rejected PJM's proposal to curtail large new loads during shortfalls, meaning a mechanism that would have rationed scarce grid capacity for data centers was struck down. This weakens M2 (special terms/pauses that tighten the power constraint) and modestly reduces the case that grid interconnection is the binding constraint, since a key rationing tool was disallowed and states retain discretion rather than a uniform federal curtailment regime being imposed.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 14 October 2026, 09:46 UTC · Main assessment
  • “Fermi Signs Credit Support Agreement with AMD and Announces Partnership” · digitaljournal.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M5 (Ways around the grid: on-site power, co-location and flexible load). Read as material.
  • On the assessment: less likely.
  • On alternative MX1: more likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: A credit support agreement between Fermi and AMD for a Texas AI campus built around securing dedicated power and compute capacity suggests on-site or co-located power arrangements are enabling campus buildout, which is a workaround to conventional grid constraints (M5). This weakens the thesis that grid power is the binding constraint. The AMD involvement also hints that accelerator supply is being bundled with the deal, slightly favouring MX1.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 14 October 2026, 09:46 UTC · The condition it rests on
  • “Fermi Signs Credit Support Agreement with AMD and Announces Partnership” · digitaljournal.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R3 (Financing for data centre campuses). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: more likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: A signed credit support agreement with AMD for a Texas campus is a concrete financing arrangement that keeps construction moving, supporting R3. The structure—a specialized provider (Fermi) with vendor credit support from AMD rather than a hyperscaler's own balance sheet—favours RX2, as spending is moving off the largest companies' balance sheets to specialized providers.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.
• Read 13 October 2026, 09:42 UTC · The condition it rests on
  • “Alphabet Signals More AI Spending After $80.6 Billion First-Half Outlay” · tokenpost.com · 13 October 2026 · Open the report
  First reported 13 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R2 (Capital spending outlook). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Alphabet's reported $80.6 billion first-half 2026 capex and signaled further increase in 2027 directly raises the capital spending outlook (R2), sustaining demand pressure on power infrastructure through 2027. This contradicts RX1's flat/declining growth scenario; it neither confirms nor denies RX2 since the report concerns Alphabet's own balance sheet spending.

  Outcome: Waiting for a second outlet to report it, until 20 October 2026.

Developments read before the record began

Before the record began, the update loop ran each day as it does now but could write nothing. The rule does not weigh a development first reported before the record began, and these readings did not set the starting point, which a separate reading set.

• Read 12 October 2026, 09:42 UTC · The condition it rests on
  • “Alphabet's Google Cloud Is Surging at 82%, and the Heavy Spending Is Starting to Make Sense” · theglobeandmail.com · 11 October 2026 · Open the report
  First reported 11 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R2 (Capital spending outlook). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Alphabet guided to $195–205 billion in 2026 capex, a substantial figure reflecting continued heavy investment in AI infrastructure, which strengthens the view that capital spending keeps rising and demand keeps pressing on power constraints. The note that supply is limiting ability to meet demand directly supports the thesis that power/interconnection is the binding constraint rather than capital. RX2 is neither favoured nor disfavoured since the report concerns Alphabet's own balance-sheet spending, not a shift to specialized providers.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 12 October 2026, 09:42 UTC · Main assessment
  • “Powering the AI revolution: Inside the grid, thermal, and interconnect challenges limiting data center growth” · techradar.com · 11 October 2026 · Open the report
  First reported 11 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M1 (Waits for large grid connections). Read as routine.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report describes grid connection waits of up to 14 years in Northern Virginia, directly strengthening M1 (long waits for large grid connections) and supporting the thesis that power/interconnection is the binding constraint. However, this is commentary/analysis rather than a regulatory order or project filing, so it is rated routine.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • Headline not shown · news.cnyes.com · 10 October 2026 · Open the report
  First reported 10 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Texas has enacted new rules requiring large electricity users (including AI data centres) to post financial guarantees and meet development milestones to secure or retain grid interconnection capacity. This is a concrete regulatory action imposing special terms—collateral and milestone requirements—on large new loads, directly matching the M2 definition and tightening power access as a binding constraint. It does not implicate chip supply, financing costs, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “The Build-Out Is Real. Only Northern Virginia’s Meters Show It. • Integrity Research” · integrity-research.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M1 (Waits for large grid connections). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report cites concrete figures: ERCOT tracking ~438 GW of large-load requests by June 2026 with the vast majority having no studies submitted, and Dominion reporting ~25 GW of requests with connection dates stretching through 2031. These are reported regulatory and utility filings showing queues far exceeding near-term grid capacity, directly evidencing long waits for grid connection as the binding constraint rather than chips, capital, or permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “Who pays if Ohio data center projects don't materialize?” · cleveland.com · 6 October 2026 · Open the report
  First reported 6 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC is actively examining PJM rules to require financial commitments from large new loads (data centers), which constitutes regulatory action to impose special terms—deposits or collateral—on new grid connections, directly rationing scarce transmission capacity and strengthening M2. This does not implicate chip supply, financing costs, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · The alternative
  • “The AI capex ledger: hyperscaler spending versus cloud revenue, 2024 to 2026 | BlockWest Research” · blockwest.co · 6 October 2026 · Open the report
  First reported 6 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Weakens factor C4 (AI revenue falling behind spending). Read as material.
  • On the assessment: less likely.
  • On alternative CX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative CX2: less likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report states combined capex of $165.1B in Q2 2026, up 87% YoY, while cloud revenue (Google Cloud $24.8B, AWS $42.2B) remains far smaller than spend—this shows spending is racing ahead of AI-attributable revenue, which weakens C4 (revenue falling behind spending is the condition, but here it means the assessment 'demand sets the pace' is less supported because spending is outrunning revenue justification). The gap between capex and cloud revenue suggests spending is not being paced by revenue earned, making the if_wrong assessment less likely.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · The condition it rests on
  • “The AI capex ledger: hyperscaler spending versus cloud revenue, 2024 to 2026 | BlockWest Research” · blockwest.co · 6 October 2026 · Open the report
  First reported 6 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R1 (Quarterly capital spending by the largest US cloud companies). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report states Microsoft, Alphabet, Amazon and Meta spent a combined $165.1B on property and equipment in Q2 2026, up 87% YoY and 3.7x the Q1 2024 level, with Q2 2026 described as the largest quarter of infrastructure spending—this directly strengthens R1 (quarterly capex by the largest US cloud companies keeps rising), supporting the thesis that demand keeps pressing on the power constraint and making RX1 (spending leveling off) less likely.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “NEWS: MISO and ISO-NE are making major moves to bolster the grid in 2027.” · energycentral.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Weakens factor M1 (Waits for large grid connections). Read as material.
  • On the assessment: less likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: MISO is proposing a fast-track review process for large loads (200 MW+), which represents a structural change aimed at reducing interconnection wait times. If grid operators are actively shortening the queue process, the binding constraint of long waits weakens, making grid power less likely to remain the sole pace-setter through 2027.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “Oracle's Wisconsin AI data center hits a Wisconsin power line roadblock” · startupfortune.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M1 (Waits for large grid connections). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: A Wisconsin regulatory review of the transmission line for Oracle's 1.3-GW campus has been restarted, with a completeness checkpoint set for October 19, directly illustrating that grid interconnection processes are delaying a large AI data centre project. This strengthens M1 (long waits for grid connections) and the main thesis, while providing no evidence that chips, capital, or local permitting are the cause of the delay.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “Biggest US Grid Suspends Plan for Data Center Power Auction” · energyconnects.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: PJM, the largest US grid operator, suspended its planned capacity procurement auction after FERC found it deeply flawed, leaving a reported 6.8-GW shortfall unresolved and cost allocation for data-center demand unsettled. This is a regulatory action that directly impedes how large new loads are connected and priced, strengthening the case that grid capacity and interconnection terms are the binding constraint rather than chip supply, capital, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · The alternative
  • “Nvidia, Broadcom shielded as AI power crunch hits chip supply chain, says Morgan Stanley” · rmb.reuters.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor C5 (Power still given as the reason for delays). Read as routine.
  • On the assessment: less likely.
  • On alternative CX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative CX2: less likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Morgan Stanley's estimate of a 32-GW net power shortfall through 2028 is analyst commentary citing power as the reason for potential deployment delays, directly reinforcing C5 (power still given as the reason for delays) and making the demand-led scenario less likely. It does not provide evidence of demand softening or falling cost-per-unit driving slower spending.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 11 October 2026, 09:43 UTC · Main assessment
  • “Nvidia, Broadcom shielded as AI power crunch hits chip supply chain, says Morgan Stanley” · rmb.reuters.com · 5 October 2026 · Open the report
  First reported 5 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M3 (Projects delayed, reduced or moved because of power). Read as routine.
  • On the assessment: more likely.
  • On alternative MX1: more likely.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The Morgan Stanley report explicitly frames a 32-GW power shortfall as the cause of potential deployment delays and downstream disruption to chip suppliers, strengthening M3 and the main thesis. It also notes the power crunch could disrupt chip supply chain orders, giving some support to MX1 (hardware supply disrupted as a downstream consequence of power constraints), though power remains the stated primary cause.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “DOE presses PJM on ratepayer protections from large load costs | Utility Dive” · utilitydive.com · 8 October 2026 · Open the report
  • “DOE Presses PJM On Ratepayer Protections From Large Load Costs” · zerohedge.com · 9 October 2026 · Open the report
  First reported 8 October 2026. Reported by 2 outlets when it was read. Outlet grade: medium, the best of them.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The DOE's formal urging of PJM to revise its large-load procurement plan so that data centers bear the costs they cause represents a regulatory intervention specifically addressing cost allocation and connection terms for large new loads, directly consistent with M2's definition of special terms rationing scarce connection capacity. This action signals that grid capacity constraints are real enough to require federal-level intervention, strengthening the case that power and interconnection are binding constraints rather than chips, capital, or local permitting.

  Disputed by the challenge step, machine-written by google/gemini-2.5-flash: The reports state that the DOE urged PJM to revise its procurement plan to prevent cost shifts onto existing utility customers from data centers, implying that data centers should bear more of the costs they cause. While this relates to cost allocation for large loads, it does not explicitly mention 'minimum-demand charges, collateral, study deposits or pauses on new connections' as defined in M2. The reports focus on preventing cost shifts, which is a different aspect than the specific 'special terms' listed in the factor definition for rationing scarce capacity.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · The condition it rests on
  • “Distress Watch: Data Center Debt Selloff Snags AI Buildouts” · news.bloomberglaw.com · 6 October 2026 · Open the report
  First reported 6 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Weakens factor R3 (Financing for data centre campuses). Read as material.
  • On the assessment: neither more nor less likely.
  • On alternative RX1: neither more nor less likely.
  • On alternative RX2: neither more nor less likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The reported debt selloff and heightened financing scrutiny for data centre campuses weakens R3 (financing closing on schedule), suggesting construction financing faces new headwinds. However, this does not directly address whether the largest cloud companies' own capital spending is rising or leveling off, nor does it clarify whether spending is shifting to specialized providers, so it does not move the assessment or its alternatives materially.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Duke Energy protects customers from data center costs” · wowktv.com · 7 October 2026 · Open the report
  • “Duke Energy strikes deal with data center giants” · axios.com · 7 October 2026 · Open the report
  First reported 7 October 2026. Reported by 2 outlets when it was read. Outlet grade: medium, the best of them.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Duke Energy and large customers have proposed a North Carolina settlement requiring major electricity users to fund dedicated facilities, provide security deposits for shared grid upgrades, and meet minimum payment obligations—exactly the kind of special financial terms and cost-allocation rules that ration scarce connection capacity. This is a concrete regulatory filing, not mere commentary, and it directly strengthens M2 by showing utilities and state regulators imposing conditions that make grid access more constrained and costly for new large loads.

  Upheld by the challenge step, machine-written by google/gemini-2.5-flash: The reports clearly state that Duke Energy and large customers have proposed a settlement in North Carolina that includes requirements for major electricity users to fund dedicated facilities, provide security deposits for shared grid upgrades, and meet minimum payment obligations. This directly aligns with the definition of Factor M2, which describes special terms for large loads, such as minimum-demand charges and collateral, that ration scarce connection capacity. The development is a concrete regulatory filing, not commentary, and directly strengthens the factor by showing utilities and state regulators imposing conditions that make grid access more constrained and costly for new large loads.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Amazon deal will enable $3 billion Calvert Cliffs investment, power data centers” · baltimoresun.com · 30 September 2026 · Open the report
  • “Amazon and Constellation sign 20-year nuclear power purchase agreement for 690 MW at Calvert Cliffs — Private Infrastructure Daily” · privateinfrastructuredaily.com · 5 October 2026 · Open the report
  First reported 30 September 2026. Reported by 2 outlets when it was read. Outlet grade: unknown, the best of them.
  Strengthens factor M5 (Ways around the grid: on-site power, co-location and flexible load). Read as material.
  • On the assessment: less likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Amazon secured 690 MW via a 20-year PPA with Constellation at Calvert Cliffs, including backing a nuclear uprate, demonstrating that co-location with an existing power plant and long-term offtake agreements can deliver large-scale power outside conventional grid upgrade timelines. This strengthens M5 (ways around the grid), suggesting power constraints can be worked around, making the thesis that grid power is the binding constraint somewhat less likely.

  Disputed by the challenge step, machine-written by google/gemini-2.5-flash: The reports confirm Amazon's PPA with Constellation for 690 MW from Calvert Cliffs, including backing an uprate. However, the reading claims this demonstrates 'co-location with an existing power plant' and that it 'can deliver large-scale power outside conventional grid upgrade timelines.' The reports do not explicitly state that Amazon's data centers will be co-located at the Calvert Cliffs site, nor do they provide information about the timeline of power delivery relative to conventional grid upgrades. Therefore, the reading makes an unsupported inference about the implications for M5.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “FERC rejects Oklo complaint seeking to reinstate project in PJM interconnection study cycle” · utilitydive.com · 25 September 2026 · Open the report
  • “FERC Backs PJM on Oklo Virginia Project, Finding No Rules Breach — EnergyReader.io” · energyreader.io · 29 September 2026 · Open the report
  First reported 25 September 2026. Reported by 2 outlets when it was read. Outlet grade: medium, the best of them.
  Strengthens factor M1 (Waits for large grid connections). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC's rejection of Oklo's complaint confirms that a 750-MW project in Virginia—a corridor with the grid's sharpest load growth—has been dropped from PJM's active interconnection study cycle, a regulatory ruling that directly extends the wait for grid connection and illustrates how interconnection queue disputes are tightening around large loads. This strengthens M1 (longer waits for large grid connections) and supports the main assessment without providing evidence for chip shortages, financing constraints, or local permitting as the binding factor.

  Disputed by the challenge step, machine-written by google/gemini-2.5-flash: The reports discuss a 750-MW 'mixed-technology project' and do not explicitly state that this project is an AI data center or directly related to AI data center buildout. While the search-engine summary for Report [1] mentions 'data-center-adjacent demand,' this is a model-written summary and not part of the report's content, making the direct link to AI data centers speculative rather than stated fact. Therefore, the development does not directly strengthen the factor M1 as it pertains to AI data centers.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “FERC orders all six US grid operators to justify or reform large electric load interconnection rules” · gcn.com · 30 September 2026 · Open the report
  • “Who Gets to Plug In: FERC's Rewrite of the Rules for Connecting Data Centers to the Grid · Inside the Datacenter” · insidethedatacenter.com · 3 September 2026 · Open the report
  • “Long Duration Energy Storage 2026 | FERC Orders RTO Reforms” · usa.long-duration-energy-storage-summit.com · 10 September 2026 · Open the report
  First reported 3 September 2026. Reported by 2 outlets when it was read. Outlet grade: unknown, the best of them.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC issued formal show-cause orders on June 18, 2026 to all six US grid operators requiring them to justify or reform their large-load interconnection tariffs (for loads over 20 MW), with operator responses due through November 2026 that will determine which campuses get energized in 2027 and on what terms. This is a concrete regulatory action imposing special scrutiny on large-load connection terms, directly strengthening M2 (special terms/tariffs for large loads) and confirming that grid interconnection rules are actively contested and potentially constraining, supporting the thesis that power and interconnection bind the buildout pace.

  Disputed by the challenge step, machine-written by google/gemini-2.5-flash: The reading states that FERC issued show-cause orders on June 18, 2026, with operator responses due through November 2026. However, all provided reports are dated September 2026, indicating they were published after the alleged FERC action but before the response deadline. The reports confirm the FERC orders were issued on June 18, 2026, but they do not state that these orders will determine which campuses get energized in 2027 or on what terms, as the reading claims. The reports only indicate that the orders require justification or reform of existing rules, not that they immediately dictate 2027 energization outcomes.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · The alternative
  • “The Pilot #73 - How to win friends and export AI” · pacforum.org · 9 October 2026 · Open the report
  First reported 9 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor C5 (Power still given as the reason for delays). Read as routine.
  • On the assessment: less likely.
  • On alternative CX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative CX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report cites two-to-four-year grid-connection delays for new data centres, directly reinforcing that power and interconnection remain binding constraints, which strengthens C5 and makes the demand-paced scenario less likely. It is commentary/analysis rather than a regulatory order or filed figure, so materiality is routine.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Map Shows States Taking Action to Protect Ratepayers from Data Center Power Costs” · newsweek.com · 9 October 2026 · Open the report
  First reported 9 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: more likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Multiple states are reported as adopting cost-allocation rules, reviews, or permitting pauses specifically weighing data-center growth against grid costs, which constitutes special regulatory terms rationing connection capacity and strengthens M2. Some pauses may also cite community impacts, which partially supports MX3, but the grid-cost framing is the primary driver cited.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Why Texas Is Making Data Centers Wait” · zerohedge.com · 9 October 2026 · Open the report
  First reported 9 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Texas has implemented a large-load study process with deposits and an audit pause, which are precisely the special terms (deposits, study requirements, connection pauses citing grid capacity) that factor M2 defines as rationing scarce connection capacity. This directly supports the thesis that grid power and interconnection are the binding constraint through 2027, while providing no evidence that chips, capital, or local permitting are the limiting factor.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Oracle Trucks in Natural Gas to Outrun Pipeline Delays at AI Data Centers” · webpronews.com · 9 October 2026 · Open the report
  First reported 9 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M5 (Ways around the grid: on-site power, co-location and flexible load). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Oracle is physically trucking natural gas to data centre sites to work around delayed pipeline and grid infrastructure, which is a concrete operational workaround confirming that power access (grid and pipeline) is the active constraint forcing creative solutions. This strengthens M5 as a real but partial mitigation while reinforcing that grid/power is the binding constraint rather than chip supply or capital.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • Headline not shown · tokenpost.kr · 9 October 2026 · Open the report
  First reported 9 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report describes PJM pursuing tariff revisions requiring large new loads to pay for the generation and grid capacity needed to serve them, which is precisely the kind of special financial term (cost-allocation tariff) that rations scarce connection capacity and makes grid power a binding constraint. This directly strengthens M2 and supports the thesis that power/interconnection, not chips or capital, is the limiting factor.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · The condition it rests on
  • “AWS CEO Matt Garman defends AI spending with a spread-out customer base” · cryptobriefing.com · 8 October 2026 · Open the report
  First reported 8 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor R1 (Quarterly capital spending by the largest US cloud companies). Read as material.
  • On the assessment: more likely.
  • On alternative RX1: less likely. This tells the alternative apart from the assessment.
  • On alternative RX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report states Amazon plans roughly $220 billion in 2026 capital spending on AI infrastructure, indicating continued large-scale spending by a major cloud company that keeps demand pressing on power constraints. This directly contradicts RX1's scenario of spending leveling off, while RX2 is neither supported nor refuted since this is AWS's own balance-sheet spending.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Florida Utilities Want Data Centers To Pass A Credit Check First” · forbes.com · 8 October 2026 · Open the report
  First reported 8 October 2026. Reported by 1 outlet when it was read. Outlet grade: medium.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as routine.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Florida utilities seeking credit checks (collateral/deposit requirements) for large data-centre loads is precisely the kind of special financial term that rations scarce connection capacity under M2, strengthening the view that power and interconnection are the binding constraint. However, no final regulatory order had been issued at publication, making this commentary/proposed action rather than a concluded regulatory change.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Democratic candidates seek probe of data centers' impact on APS bills” · rmb.reuters.com · 8 October 2026 · Open the report
  First reported 8 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: APS is proposing to assign new power-purchase and infrastructure costs directly to large customers (data centres), which constitutes special tariff terms for large loads—a form of rationing that strengthens M2. This bears on grid power cost allocation rather than chip supply, capital availability, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Data centers face a sweeping new power regime” · axios.com · 8 October 2026 · Open the report
  First reported 8 October 2026. Reported by 1 outlet when it was read. Outlet grade: medium.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as routine.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The reported Senate proposal would require large data centres to cover grid costs they cause and could allow regulators to impose additional charges or electricity-use limits, directly expanding the regime of special terms for large loads described under M2. However, the report notes legislative action is uncertain, making this commentary/proposal rather than an enacted rule, so materiality is routine. It does not implicate chip supply, financing costs, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “FERC Accepts and Suspends PJM’s Reliability Backstop Procurement Proposal, Establishes Paper Hearing Procedures, and Encourages a New Section 205 Filing | Troutman Pepper Locke - JDSupra” · jdsupra.com · 7 October 2026 · Open the report
  First reported 7 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: FERC's acceptance of PJM's Reliability Backstop Procurement proposal and the establishment of a proceeding examining LSE collateral requirements directly reflects utilities and regulators imposing special financial terms (collateral requirements) on large new loads, consistent with M2. This is a regulatory order, not mere commentary, and it signals grid capacity constraints driving new interconnection conditions rather than chip supply, financing costs, or local permitting issues.

  Disputed by the challenge step, machine-written by google/gemini-2.5-flash: The reports mention 'LSE collateral requirements' but do not explicitly state that these requirements are being imposed on 'large new loads' or specifically on AI data centers. The factor M2 is defined by 'special terms for large new loads,' and without this explicit connection, the development does not directly strengthen the factor as defined. The reports do not provide enough detail to confirm that these collateral requirements are a direct response to grid capacity constraints caused by AI data center buildout.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · The alternative
  • “AI Data Centers Hit a "Power Crunch": A Full Breakdown of the Electricity Gap and Which Semiconductor Stocks Face the Biggest Risk?” · odaily.news · 7 October 2026 · Open the report
  First reported 7 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor C5 (Power still given as the reason for delays). Read as routine.
  • On the assessment: less likely.
  • On alternative CX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative CX2: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report cites power and grid-connection delays as limits on accelerator deployment and highlights large interconnection queues, directly reinforcing C5 (power still given as the reason for delays), which weakens the if_wrong scenario. The report is commentary/analysis rather than a material regulatory order or project filing.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “How Communities Are Halting the AI Power Grab” · webpronews.com · 7 October 2026 · Open the report
  First reported 7 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: more likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: The report describes a Texas review of large-load interconnections, which constitutes a regulatory pause on new connections citing grid capacity, directly strengthening M2. However, the report also describes local opposition and permitting delays, which simultaneously supports MX3, so both the main assessment and MX3 are favoured by this development.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · Main assessment
  • “Duke Energy Corporation - Duke Energy protects customers from data center costs” · investors.duke-energy.com · 7 October 2026 · Open the report
  First reported 7 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Strengthens factor M2 (Special terms for large loads: tariffs, deposits and pauses). Read as material.
  • On the assessment: more likely.
  • On alternative MX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX2: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative MX3: neither more nor less likely. This tells the alternative apart from the assessment.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Duke Energy reached a formal settlement in North Carolina imposing special financial terms (cost-to-serve protections, deposits/charges) on large-load customers including data centres, which is precisely the kind of regulatory mechanism that rations scarce connection capacity and makes grid power a binding constraint. This does not implicate chip supply, financing costs, or local permitting.

  Outcome: First reported before the record began, so the rule does not weigh it.
• Read 10 October 2026, 09:45 UTC · The alternative
  • “Cipher Mining Sees AI Demand Holding Strong as Power Scarcity Drives Data Center Leases” · marketbeat.com · 6 October 2026 · Open the report
  First reported 6 October 2026. Reported by 1 outlet when it was read. Outlet grade: unknown.
  Weakens factor C5 (Power still given as the reason for delays). Read as routine.
  • On the assessment: less likely.
  • On alternative CX1: neither more nor less likely. This tells the alternative apart from the assessment.
  • On alternative CX2: less likely.

  Its reason, machine-written by anthropic/claude-sonnet-4.6: Cipher Mining's CFO states that scarce power remains the main constraint for AI data-center tenants, directly reinforcing that the grid still binds and weakening the demand-paced scenario. This is commentary from a company executive rather than a regulatory order or filed figure, making it routine.

  Outcome: First reported before the record began, so the rule does not weigh it.

Where the sealed record is

The numbers, the changes and their fingerprints are on the judgment’s page and the changes page. How the starting point was read, and the rule the loop applies to developments first reported before the record began, are on the method page.