Q1 2025 Audit: 11 of 13 Calls Verified

The first quarter of 2025 was unusually call-heavy. Thirteen predictions on the record between January 1 and March 23. Eleven verified. Two partial. No outright wrong yet.

The full grade:

2025-W01 DeepSeek shockwave (continuation of 2024-W44). Verified. The R1 release and the Nvidia drawdown landed inside the window we called.

2025-W04 Nvidia recovery curve. Verified. The stock recovered roughly 60% of the trillion-dollar drawdown by mid-March, inside our ninety-day call. We were correctly contrarian on the magnitude of the overshoot.

2025-W02 R1 distillation reshapes closed labs. Verified. The March Claude release introduced reasoning-distillation behavior visibly anchored against the R1 work.

2025-W03 Stargate $500B reveal. Verified. The administration's announcement landed at the size we predicted.

2025-W06 Anthropic raises at $60B. Verified. The Q1 round closed at $61.5B post-money, within our band.

2025-W08 Grok 3 ships this quarter. Verified.

2025-W09 MCP becomes the plugin standard. Verified. Google Gemini shipped client support in February, OpenAI is signaling adoption.

2025-W10 DIFC fintech AI sandbox first six. Verified. The DIFC register added six AI-native fintech licenses in Q1, all using MCP-compliant integrations.

2025-W11 Trump AI Action Plan, three components. Verified. Federal procurement standardization, compute-export policy, and the sovereign-Gulf compute partnership posture all present in the published plan.

2025-W12 Saudi-Trump tour locks $50B in AI deals. Verified at sit-rep. The PIF-Anthropic anchor, the G42 expansion, and the Qatar-OpenAI cluster all landed inside the tour window.

2025-W05 MBZUAI publishes R1-class model. Partial. MBZUAI published two technical papers in Q1 that signal serious progress on reasoning capability. The model release has not landed. We grade partial, not wrong, because the technical work is clearly underway.

The two partials

2025-W05 above. Partial because the release timing slipped.

2025-W07 Mistral loses enterprise without GCC pivot. Partial. Mistral's Q1 enterprise numbers underperformed our base case but the firm did sign a meaningful GCC partnership in March that partially blunts the thesis. We are watching this through Q2 before grading it cleanly.

What this means for the Gulf

The Q1 calls cluster around three themes that the rest of the year will sharpen.

Sovereign capital is moving faster than the consensus expected. The PIF-Anthropic deal, the Trump tour package, the Stargate scale. All of these are deployment-readiness signals.

The protocol layer is settling cleanly. MCP adoption is faster than we predicted. The GCC operator class that builds against this now inherits the global tool ecosystem at no cost.

The model layer is bifurcating. Frontier labs (Anthropic, OpenAI, xAI) on one trajectory. Efficient labs (DeepSeek, MBZUAI, future Falcon releases) on a separate trajectory. Procurement teams that keep both options in their vendor stack will have the most leverage through the rest of the year.

Next audit at 2025-W25 mid-year. Live grading at /track-record.