AI SYSTEMIC FINANCING RISK · CANARY DASHBOARD

Should I act?

Watch the canaries in AI financing: where stress starts, whether it is spreading, and when financial warning signs become a portfolio action signal. The aim is not to predict a fixed crash date, but to track how the evidence and the decision evolve.

Research prototype · financial evidence reviewed through November 5, 2026 (close plus post-close issuer disclosures) · source dates retained individually · risk scores and forecasts are not calibrated probabilities

AS OF NOVEMBER 5, 2026 · MODEL SIGNAL Loading SFR decision… Checking the model's credit, trend and breadth confirmation gates. See the escalation stage and what would change this

November 5, 2026 · regular-session closing review

AI rout confirmed at the close: Nasdaq −5.52%

Nasdaq Composite: 24,315.00, down 1,421.52 points. The −5.52% return is calculated from AP's closing figures (reported as −5.5% rounded). AP closing summary ↗

The index ended about 11.9% below its October high, erasing Wednesday's rebound. AP closing report ↗

Reuters separately reports renewed CoreWeave CDS stress, with no fresh numerical closing spread verified. Reuters via Mint ↗ The earlier credit story is superseded at the same analyst weight. This does not establish broad credit contagion or two financing failures.

A Reuters report now corroborates the semiconductor index's intraday fall of as much as 11%; it is not a closing return. Reuters sector report ↗ IREN and Nebius closing returns are now verified in the full-cohort table below. Old CDS and capitalization sensitivity inputs are dated proxies.

Tomorrow: conditional outlook

Scheduled catalyst: BLS calendar ↗

No numerical Nasdaq-return target or scenario probabilities are assigned. Jobs outcomes and the next session's market response remain unknown. This outlook does not change source confirmations or execute trades.

What was refreshed—and what is still missing

Complete refresh review and source register ↗

Coverage and remaining data limits

Panel	Review treatment

All 23 companies: closing prices and latest reviewed evidence

Company	Closing price / exchange session	Latest reviewed evidence / original source date

Market capitalization access was unavailable; a share-price change is not a verified market-cap change. Credit quotes and contract balances remain incomplete. This is a full-panel review with explicit gaps, not a complete live financial feed.

November 3 midterms · buildout and financing watch

After the blue tsunami: can new data centers get built?

POLICY WATCH ADDEDThe election result alone does not satisfy the model's financial action gates. Check the action signal ↓

Evidence to collect:

Different effects across companies:

Project-status reference ↗

Local approval, power, delivery and financing: four paths to verify

Senate timing reference ↗

SFR Decision Dashboard · What to do now

Is AI financing stress escalating—and should I act?

The top line separates financial warning signs from a confirmed portfolio action signal. Do not mistake hypothetical company-stress dates for sell timing.

EVIDENCE THROUGH —

Act now? · SFR-triggered equity sale

Loading evidence…

—

——

—

What could change this soon?

Next catalyst loading…

—

— —

Where we are in the escalation process—

Company stress can be elevated even when the evidence required for a portfolio de-risk review is incomplete.

What is still needed before a de-risk review?

—

Next 7 / 14 / 30 days

Modeled vulnerability, not probability of a selloff. A scheduled earnings event can surprise in either direction.

Next dated events to check

Additional financing watch items without a confirmed date

Action criteria and evidence verification

Stage 1 needs a confirmed persistent market downtrend and broad participation in the decline, in addition to independent issuer credit stress. Stage 2 needs actual financing failures, broader credit deterioration and repeated demand misses.

View all verification signals and source status

What-if: demonstrate how the action stage changes Hypothetical inputs are not observed facts and cannot trigger trades.

Dated evidence → dynamic 9-company forecast

How quickly could systemic AI financing stress spread?

Track absolute vulnerability scores across 23 companies, the projected date when 6 or 9 cross a warning threshold, and modeled direct market-value losses. The charts run from history through the ninth affected company; the original March 2027 stress test remains a comparison.

AS OF —

Composite vulnerability · selected cohort—Absolute model index: 0–100, not default probability

Companies in Warning or Crisis (≥55)—Current evidence-adjusted score count

Modeled DIRECT loss at ninth-company horizon—Historical proxy + forecast · NOT realized loss

Six- and nine-company stress milestones—Sixth milestone and highlighted crossing

Stable 0–33 Watch 34–54 Warning 55–76 Crisis 77–100

Vulnerability · daily history and forecast0–100 model score · risk zones

Oct 14, 2026Forecast to ninth-company threshold

Daily changes and future forecast share one chart. Historical change bars use a separate labeled scale.

Direct market-value lossUSD trillions · historical proxy + forecast

Historical portion is model-implied, NOT actual historical market-value losses

Latest evidence strengthening the risk thesis

Latest evidence weakening the risk thesis

Historical vulnerability is an analyst-defined evidence index, not an observed credit rating. Historical direct losses are model-implied capitalization proxies, NOT actual market declines: dated issuer capitalization histories are not yet available for an observed loss series. Future dashed lines are conditional projections. The time axis expands the short historical segment and marks the change in scale. Both charts extend to the ninth company's selected threshold crossing, or 78 weeks if nine never cross. No indirect or unexplained losses. Methodology.

Oct 27

Include announced/conditional transactions

Select a weekEvidence date → breadth endpoint

TodayProjected endpoint

Projected warning breadth (conditional, not observed)0%

Stable index <34 Watch 34–54 Warning 55–76 Crisis ≥77 Investments Cloud Backstops

Circle area changes with scenario drawdown, and color is a composite modeled vulnerability index (0–100), NOT a credit rating or observed probability of failure. Network shows actors and aggregated original signed/committed links (current funded balances unreconciled); are in hypothetical modeled crisis status.

Forward Canary Timeline · evidence cutoff October 27 for original March comparison

When could financial stress spread to 6 or 9 companies?

Forward mode projects weekly vulnerability from the latest dated evidence until a chosen number of companies first crosses a stress threshold. Switch to Jay's March reference mode to examine the original reverse-stress scenario. Neither path is statistically calibrated.

ROLLING FIRST-PASSAGE MODEL · UNCALIBRATED

Reverse time axis

Forecast checkpoints

Select a projected checkpoint or press Play to explore the financial network. Today's evidence is known; all future weeks are judgment-based projections. The March stress test is a separate option.

Canary Watch — Warning Signs & Evidence

1 · What must happen for the crash path to remain credible?

2 · Leading indicators and screening thresholds

Thresholds flag questions to investigate; they are not calibrated probabilities.

3 · What we can actually verify

4 · What would weaken this crash thesis?

How stress would reach the next stage

As-of evidence is not a future observation. Network stress scores and circle changes are scenario outputs, not measured default probabilities.

Capital-market review · November 5, 2026 · individual source dates below

AI financing diverges: selective distress, major deals still priced

Dated filings and reporting separate equity market stress, actual credit facilities, priced share offerings, committed SPV loans and deals still only in talks. Funding access is uneven; the crash scenario is not inevitable.

Accounting discipline: A filed IPO is not IPO proceeds. Meta's $37.5B priced share offerings (including common forwards) are not immediate $37.5B cash receipts, and the proposed additional $60B ATM is not issuance. The Broadcom/Anthropic $63B committed SPV is not $63B drawn; its first lien, second lien and residual-value backstop are different instruments. SpaceX's binding GPU lease and Nvidia/AMD planned Meta purchases require settlement. These headlines are not added to the $20T market-loss buckets. New announced SPV debt, Broadcom's guarantee, and supplier-to-Meta equity arrows are conditional and disabled by default. Nebius's secured $3.5B facility is recorded as committed credit, not necessarily disbursed.

Market-access review · November 5, 2026 · historical IPO events dated below

Neocloud IPO / financing appetite canaries

November 5 review: OpenAI funding-target cuts and Nscale customer optionality are included in the financing panel above; earlier IPO events remain historical records. An IPO delay is not a debt default; a falling share price is not a cash outflow. All signals belong to one correlated market-access cluster.

Elice also raised ~$150M and reported operating profitability, despite its first-day share decline. These events are not additional financing arrows or realized market losses in the $20T accounting.

Direct market-value accounting · hypothetical

Direct market impact · modeled, not observed

—Seven disjoint sector buckets
— of original direct-loss reference

Comparison:

Edit scenario capitalization bases ($ trillions)

Mag 7 aggregate Nvidia (within Mag 7) Semis ex-Mag7 Other infrastructure Private labs (ex-SpaceX) SpaceX public market Neocloud listed bucket

Proxies (especially non-Mag7 semis, other infrastructure, private labs, and neocloud aggregate) are editable analyst assumptions, not authoritative full-sector market capitalization. The sector bases retain their original October 2026 reference date; current market capitalizations were unavailable. These disjoint buckets are illustrative and exclude indirect investment write-down double counting.

Calibration status · no arbitrary 3% prior

What is—and isn't—calibrated?

The rolling stress dates are sensitivity calculations using published assumptions, not empirically calibrated event-time probabilities. Likewise, the archived March reference scenario has no validated crash probability. Current closing prices do not recalibrate these assumptions.

AI-equity correction ≥20%Not calibrated

Needs dated returns and option-implied downside distributions across a defined AI stock universe.

Systemic AI financing distressNot calibrated

Needs cross-issuer credit spreads, project defaults, liquidity and debt maturity evidence; not just one issuer.

Exactly $20T by March 2027Not calibrated

Needs issuer-level valuations without overlap, correlated shock modeling, contagion calibration and timing validation.

What current evidence means

Stress observed

November 5 closes verified across all public cohort issuers; CoreWeave credit stress reported; Microsoft/neocloud procurement and IREN delivery timing reviewed. Source dates and limitations appear in the coverage table.

Counterevidence observed

IREN reports funded contracted growth despite delayed delivery; AMD reports strong data-center demand. Oracle and Meta offerings priced, and Google project support was disclosed. Pricing, commitments, draws and settled cash remain distinct.

Systemic confirmation missing

A complete independent funding-failure census, comparable broad credit series and persistent advance/decline breadth have not been verified. Unknown is not evidence that failures do not exist. The original March comparison remains an archived hypothetical scenario.

These signals are qualitative and partly correlated; no automatic likelihood update or hidden numerical prior is used. The separate CoreWeave CDS calculator estimates an issuer-specific, risk-neutral credit proxy—not the crash probability.

Counterparty inspector

Primary canary for this actor

Exposure registry · illustrative nominal dollars

Selected company flows

Click a source link to inspect disclosed status. A commitment is not revenue, cash, debt or a current loss.