Khamenei personally rejecting — rather than delegating to Pezeshkian/Araghchi — is the consequential escalation signal. Supreme-Leader-level rejection is far harder to walk back than ministerial defiance. The Sunday revised-offer window (yesterday's 55% prediction) effectively resolves negative.

The Istanbul European-track talks are the partial counter-signal: Iran is keeping a diplomatic channel open with the E3 (UK/France/Germany) even while rejecting the US bilateral track. The 'irreversible' UN-sanctions-snapback warning is Tehran's leverage threat against the European track collapsing too.

The confirmed-extension-with-continued-strikes pattern is the operative reality: the diplomatic framework holds while the kinetic violations continue. The two-track structure (State Dept political June 2-3 + Pentagon military May 29) is the most structured Israel-Lebanon negotiation architecture of the cycle — a meaningful upgrade from the prior ad-hoc extensions.

Yesterday's prediction (45-day extension holds without major violation >10 killed/day for 30 days, 60%) faces immediate stress: Saturday's 5-village strikes + 9-village displacement orders. If Saturday's casualty count stays below 10, the prediction holds; if it crosses, the threshold breaks on day 1.

The deep-strike footage is Zelensky's 'justified response' messaging after the 24-killed Kyiv apartment strike — demonstrating Ukraine's intact long-range capability targeting the Russian military-logistics and air-defense backbone (Pantsir/Tor systems are the layered air-defense Russia uses to protect rear areas).

The asymmetry continues: Russia's mass-drone tempo targets Ukrainian civilian/energy infrastructure; Ukraine's deep strikes target Russian military-logistics + oil-industry. Both are escalating in parallel. The Patriot-interceptor depletion crisis remains the binding Ukrainian vulnerability.

The 'Warsh dovish-leaning but Fed holds steady' consensus is the operative framing — it reconciles the political pressure (Trump wants cuts) with the data reality (sticky inflation + Iran oil premium make cuts indefensible). The September rate-hike probability is the surprising tail that markets are starting to price.

Warsh inherits the West-Asia-crisis uncertainty (Khamenei's Saturday rejection re-inflates the oil premium) on top of sticky inflation. The 'rougher-than-normal opening phase' framing from strategists looks validated by the Khamenei development.

The H200-clearance walk-back is the consequential market-correction signal. Thursday's records were partly built on the Reuters H200-clearance report; Saturday's 'no Nvidia breakthrough' analysis explains Friday's hard selloff and means the AI-capex-China-tailwind thesis from Thursday was premature.

Methodological note for the prediction track: Thursday's '70% H200 structurally holds' prediction needs major downward revision — the clearance may have been overstated in initial single-source (Reuters) reporting before multi-source confirmation. This is the cleanest example this cycle of why single-source breakthrough claims need 48-hour multi-source verification.

The walk-back reframes the entire Thursday-Friday market action: Thursday's records were partly an H200-clearance-driven overshoot; Friday's selloff was the correction; Saturday's analysis confirms the clearance was not a confirmed summit deliverable. The AI-capex-China-tailwind narrative loses its concrete anchor.

Strategic implication: the US AI-export-control framework remains substantively intact. Anthropic's tri-cloud + Google-TPU diversification (the Friday $200B commitment) looks prescient — it de-risks against exactly this Nvidia-China-supply uncertainty. The decoupling thesis is not dead; it was prematurely declared dead Thursday.

The OpenAI -2.9% April business-adoption decline (not just Anthropic's gain) is the consequential detail — it's an absolute decline, not just relative. The OpenAI-EU-pre-release controversy + the late-April Anthropic capacity-recovery + the PwC Big-Four-channel win all compounded into actual OpenAI enterprise share loss.

Google I/O May 19 is the next AI-vendor-cadence event. Watch whether Google leverages the Anthropic-$200B-TPU-commitment relationship narrative or counter-positions Gemini aggressively against both Anthropic and OpenAI. The infrastructure-partner-vs-model-competitor tension is the operative Google strategic question.

The Coastal Flood Advisory for San Pablo Bay (north of the Bay Bridge — Marin, Vallejo, Richmond shoreline) is the operative local-impact item. Low-lying shoreline parking, marinas, and tidal-area infrastructure face periodic minor flooding through Tuesday early morning.

Strong NW winds 20-30 mph affect Bay sailing/recreational boating and may produce localized power-line/tree issues in the East Bay and Peninsula hills. No major civic-event disruption flagged through Monday.

Google I/O's framing is the operative Bay Area question: does Google position Gemini aggressively against Anthropic/OpenAI while being Anthropic's largest TPU customer? The coopetition tension is structurally novel and will shape Bay Area AI-talent and partnership dynamics through H2.

The H200 walk-back means the Bay Area chip-stock cohort (Nvidia, AMD, Micron) opens Monday without the China-tailwind narrative. The AI-capex-floor thesis holds on domestic hyperscaler demand, but the multiple-expansion catalyst from Thursday is removed.

The official Andaman declaration converts the forecast into a confirmed event. Combined with the >110% LPA May rainfall outlook, the 2026 monsoon is now firmly tracking as one of the earliest and most favorable in two decades — a structural macro-positive for India's H2.

The heatwave's natural ending window is now defined by the May-26 Kerala onset + northward progression. The isolated final-phase heat over Rajasthan/Gujarat/MP/Maharashtra is the last acute mortality window; most of mainland India's heat risk closes by mid-June.

The ~37-MLA figure (up from the ~30-31 tracked Friday) confirms the Shanmugam-Velumani faction has consolidated decisively. EPS's faction is now a ~10-MLA rump — far below the 1/3 threshold needed to contest the split exemption. The anti-defection disqualification risk for the majority is effectively eliminated.

The operative remaining fight is the EC 'two leaves' party-symbol allocation (a 3-6 month adjudication) — which faction is legally 'the AIADMK.' The 37-vs-10 numerical reality strongly favors Shanmugam-Velumani on the legislature-group test, though EC weighs organizational-wing control separately.

The downgrade window is open NOW but is condition-dependent — it only benefits cases where the EB-3 priority date is meaningfully ahead AND the underlying job qualifies for EB-3 classification. Practitioners advising case-by-case evaluation rather than blanket downgrade.

The Sept-30 cliff is the structural risk: if DOS makes EB-2/EB-1 India 'unavailable' for the FY2026 remainder, all in-pipeline AOS cases freeze until October. The weekend messaging emphasizes filing everything filable now ahead of any cliff.

The narrow protected exception remains: SIJS beneficiaries whose petitions were approved April 7-June 6, 2025, or filed before May 1, retain deferred-action eligibility under the pre-rescission Policy Manual. Practitioners verifying client filing dates against these windows.

The multi-state worker-departure pattern (CA/FL/TX/AZ/NV/OR documented over the past week) strengthens the plaintiffs' factual record. The nationwide-class scope means any affected-category worker is a potential class member.

Pairs cleanly with the Airbnb CEO's earnings-call statement May 7 ('people who have lots of recurring one-on-ones are not going to survive').

Runtime: 70 minutes.

Reich frames 'Bronze Age inflection' on three concrete genetic-evidence pillars: rapid population replacement velocity, geographic specificity of replacement patterns, and linguistic lineage tracing.

Discovery pick — outside the usual AI/tech lens. Within the 14-day window through May 22; will rotate out after.

The 'jagged frontier' framing — capability gaps move outward unevenly, with sharp jumps in some domains and stalls in others.

Runtime: ~95 minutes. Within the 14-day window through May 19.

Methodology note: this is the cycle's clearest example of incremental-signal-tracking working — the prediction moved 65→55 over 24 hours on Araghchi's distrust framing, then resolved negative on Khamenei's personal rejection. The hedge direction was right even though the binary resolved against the headline number.

Forward prediction: probability of US strike-package authorization within the May 18-22 window now rises to 45% (from the prior 20-30% range). Khamenei's personal rejection removes the diplomatic-progress off-ramp; the Istanbul E3 track is the only remaining channel.

Why down 40pp: the prediction was built on the Thursday Reuters report. Saturday's multi-source analysis indicates the clearance was premature single-source reporting, not a confirmed summit deliverable. This is the cleanest cycle example of why single-source breakthrough claims need 48-hour multi-source verification before high-confidence prediction.

The residual 30% accounts for the possibility that BIS case-by-case licensing proceeds quietly even without a summit-headline confirmation — Nvidia could still get individual licenses approved without a formal announcement. But the structural-tailwind thesis from Thursday is substantially discredited.

Why up 3pp: the ~37-MLA figure is decisively past the 32-MLA threshold with margin. The structural math is now near-deterministic — anti-defection law cannot apply to a faction that constitutes >2/3 of the original party's legislature group.

Residual 7%: pure procedural risk — a TVK-aligned-Speaker adjudication challenged in Madras HC producing a timeline-extending stay. The substantive outcome (no disqualification) is near-certain regardless of procedural path.

Accuracy note: the digest's morning card referred to 'Khamenei' without distinguishing — this evening card corrects to Mojtaba Khamenei, the current Supreme Leader. The substantive point holds (Supreme-Leader-level rejection, hardest line of the cycle) but the named individual is the son, in power since March.

Mojtaba Khamenei's 'keep Hormuz closed + stop hosting US bases' framing is more maximalist than the Pezeshkian/Araghchi negotiating posture. It signals the closure is a Supreme-Leader-held lever, not a negotiable ministerial position — raising the strike-package probability for the May 18-22 window.

The Saturday casualty count (3 killed) stays below the >10-killed-single-day threshold in yesterday's prediction (45-day extension holds without major violation for 30 days, 60%). The prediction survives day 1 of the extension — small-scale-daily-violations are inside the model; a >10-killed day would break it.

The ~3,000 cumulative death toll is the structural scale that makes the State-Dept-June-2-3 / Pentagon-May-29 two-track talks consequential. The paramedic killing specifically (medical-personnel targeting) is the kind of incident that historically triggers UN/ICRC formal protest — watch for that next week.

The 91% interception rate is the consequential signal — it suggests either reduced Russian drone volume OR improved Ukrainian air-defense replenishment (or both). The earlier-week Patriot-depletion-crisis trajectory may have eased if interceptor transfer occurred over the weekend (unconfirmed).

294 drones is a meaningful step down from the 600-800-daily peak (May 13-14, the period that produced the 24-killed Kyiv apartment strike). Whether this is a deliberate Russian tempo-moderation or a resupply-cycle pause determines the next-week trajectory.

WTI's ~10% weekly gain is the cleanest market signal of the Iran-hardening. The IEA's ~4M-bpd Hormuz-flow-reduction figure is the structural supply-shock quantification — that's roughly 4% of global supply disrupted, consistent with the sustained $100+ Brent regime.

Implication for Warsh's first FOMC (June 16-17): the oil-driven inflation overhang + the falling 2026-rate-cut odds + the rising September-rate-HIKE probability collectively make Warsh's first-meeting communication challenge materially harder than at confirmation. The morning prediction (Warsh first-week hawkish-on-inflation, ambiguous-on-cuts, 70%) is reinforced by the weekly oil action.

The Amodei-White-House meeting is the consequential escalation: it means Mythos's offensive-cyber-reasoning capability is being treated as an executive-branch national-security matter, not just a congressional-briefing topic. This is the policy-engagement equivalent of the controlled-access strategy paying off — Anthropic is shaping the regulatory framework rather than reacting to it.

Strategic contrast with OpenAI: OpenAI's EU-pre-release approach generated DC pushback; Anthropic's holdback + direct-White-House-engagement generates policy-shaping access. The two vendors' regulatory strategies have diverged into a clear A/B test that resolves over H2 2026.

The strategic logic: by being the vendor that proactively brings its most-capable cybersecurity model to the White House for risk discussion, Anthropic positions itself as the trusted-partner-of-record for any future federal frontier-AI-security framework. This is the policy-channel equivalent of the PwC Big-Four-distribution moat.

Counter-pattern risk: deep executive-branch entanglement cuts both ways. If a future administration views Anthropic's policy-shaping access as regulatory capture, the relationship becomes a liability. The strategy is high-reward but increases political-exposure beta.

Google's I/O framing options: (a) lean into the Anthropic-$200B-TPU relationship as validation of Google Cloud/TPU infrastructure leadership, or (b) counter-position Gemini aggressively as the integrated alternative to both Anthropic and OpenAI. Likely a hybrid — infrastructure-partner narrative for enterprise + Gemini-competitive push for consumer/developer.

Watch for: Gemini enterprise-tier announcements (direct Anthropic/OpenAI business-share counter), any TPU-capacity-expansion framing tied to the Anthropic deal, and agentic-coding tooling (the segment where Anthropic's Claude Code leads).

North Bay shoreline residents/boaters: the periodic minor tidal flooding continues through early Tuesday. No major civic disruption flagged. Strong NW winds may produce localized East Bay/Peninsula-hills power-line issues.

Google I/O week brings elevated Mountain View/Shoreline traffic and Caltrain demand May 19-21. The conference's AI-vendor-positioning content (Gemini vs Anthropic/OpenAI) is the operative Bay Area tech-narrative driver for the week.

The talent-flow inversion is the operative Bay Area structural signal. Compensation packages at both companies will adjust within 30-60 days as the recruiting dynamics shift. The October-IPO talk specifically accelerates senior-talent interest in Anthropic (pre-IPO equity upside).

Counter-pattern: OpenAI retains scale advantages (consumer reach, ChatGPT brand, Microsoft-Azure depth). The inversion is in enterprise-business-adoption + policy-positioning + capital-narrative, not consumer or raw research headcount. The two-vendor competition is reconfiguring, not resolving.

The 'officially split' confirmation moves the morning prediction (93% Shanmugam-Velumani escapes Tenth Schedule) to effectively resolved. The remaining contest is the EC 'two leaves' party-symbol allocation — a 3-6 month adjudication where the 37-vs-10 legislature-group reality strongly favors Velumani-Shanmugam, though EC weighs organizational-wing control separately.

The TVK government's 144-MLA majority is structurally insulated regardless of the AIADMK internal outcome. The operative forward question shifts to whether the Velumani-Shanmugam faction formally merges into TVK or remains a distinct allied party — which determines TN's 2031-cycle opposition structure.

The May 16-21 northern-heatwave-intensification is the last acute mortality window before the monsoon's northward progression. The Punjab/Haryana/Chandigarh agricultural belt (wheat-procurement + early-kharif-prep) faces the most economically consequential heat stress.

Counter-balancing positive: the confirmed early monsoon (Kerala May 26, ±4d) means the northern heatwave has a defined endpoint. The monsoon's typical northward progression reaches the northern plains by late June — the heat window is acute but bounded.

The EB-3-ahead-of-EB-2 inversion is condition-dependent: it only benefits cases where the EB-3 priority date is meaningfully ahead AND the underlying role qualifies for EB-3 classification. Blanket downgrade is not the advice — case-by-case evaluation is.

The Sept-30 cliff is the structural risk: a DOS 'unavailable' designation for the FY2026 remainder freezes all in-pipeline AOS cases until the October FY2027 reset. Weekend practitioner messaging consistently emphasizes filing everything filable now.

The Senate Judiciary scheduling (Monday) is the parallel political-pressure track to the EDNY litigation. Bipartisan procedural framework forming — the hearing would formalize the three-pronged-narrowing pattern as a congressional-oversight matter.

The narrow protected exception holds: SIJS beneficiaries with petitions approved April 7-June 6, 2025 or filed before May 1 retain pre-rescission Policy Manual eligibility. Practitioners continue verifying client filing dates against these windows.

Pairs cleanly with the Airbnb CEO's earnings-call statement May 7 ('people who have lots of recurring one-on-ones are not going to survive').

Runtime: 70 minutes.

Reich frames 'Bronze Age inflection' on three concrete genetic-evidence pillars: rapid population replacement velocity, geographic specificity of replacement patterns, and linguistic lineage tracing.

Discovery pick — outside the usual AI/tech lens. Within the 14-day window through May 22; will rotate out after.

The 'jagged frontier' framing — capability gaps move outward unevenly, with sharp jumps in some domains and stalls in others.

Runtime: ~95 minutes. Within the 14-day window through May 19.

Why 45% (not higher): Trump's 'losing patience' is rhetorical-escalation consistent with the prior deferral pattern (Tuesday's CENTCOM briefing produced no authorization). Pentagon munitions-shortage concerns remain a face-saving deferral lever. The E3 Istanbul track provides a non-kinetic alternative.

Why 45% (not lower): Supreme-Leader-direct rejection + the June 30 War Powers deadline approaching + the sustained $100+ oil regime create cumulative escalation pressure. A limited 'self-defense strike' (like May 14's port strikes) is more probable than the full 'short and powerful' package.

Why 95% (effectively resolved): the 'officially split' confirmation + ~37-MLA consolidation makes the structural math near-deterministic. The only residual risk is procedural (a Madras HC stay on Speaker adjudication extending the timeline) — the substantive outcome is settled.

Forward prediction shifts to the EC 'two leaves' party-symbol allocation: probability the Velumani-Shanmugam faction is recognized as the legitimate AIADMK (or retains symbol-claim parity) within 6 months — initial estimate 55%, given EC weighs organizational-wing control separately from legislature-group strength.

Why unchanged: the day-1 outcome (3 killed) is exactly the small-scale-violation pattern the 60% probability assumed. No new information shifts the base rate.

The operative stress windows: the next two weekends (May 23-24, May 30-31) and the May-29 Pentagon military-track open. The May 9-10 precedent (~50 killed despite active ceasefire) shows the >10-threshold can break on a single bad weekend.

Why unchanged: the weekly oil action reinforces rather than alters the prediction. A market facing a ~10% WTI weekly gain + sticky CPI/PPI needs a credible Fed inflation-containment signal; Warsh's narrowest-ever confirmation margin makes the hawkish-credibility-buy the rational first move.

Residual 30%: Trump's weekend rate-cut pressure intensifies; Warsh surprise-doves to align politically, accepting the 'caving' narrative. The oil overhang makes this scenario less likely than a week ago but it remains a real ~30%.