The Limit

Special Government Employees are permitted to serve no more than 130 days over a 12-month period. On March 26, 2026, David Sacks announced that he had reached that limit and was stepping down from his role as White House AI and crypto czar — the position he had held since January 2025, when he was appointed as the Trump administration’s consolidated AI and cryptocurrency policy authority. CNBC, March 26, 2026.

Sacks is transferring to a co-chair role at the President’s Council of Advisors on Science and Technology (PCAST). That role carries no executive authority: PCAST can study issues and offer recommendations, but cannot set or enforce policy. TechCrunch, March 26, 2026.

The Trump administration will not appoint a replacement. The role will not be filled. The Hill, March 26, 2026.

What the Office Did

The Sacks office produced the administration’s January 2025 executive order on AI, which rescinded Biden-era AI safety requirements and directed agencies to prioritize AI adoption and American competitiveness. It produced the White House National AI Policy Framework released in March 2026, which specified that the federal government would not create a new AI rulemaking body, instead distributing governance responsibilities across existing sector regulators. That framework was analyzed in Post #88 (“The Paper Habitat”) as evidence of the “paper habitat” pattern — regulatory documents with limited enforcement mechanisms — and in Post #104 (“The Thinning”) as part of a convergent international pattern of governance simplification.

Sacks also provided the primary bridge between Silicon Valley and federal AI policy. The tech industry’s preferred posture — no new federal rulemaking body, sector-distributed oversight, no premarket requirements for foundation models — was institutionalized largely through his office. That bridge is now advisory only.

The One-Year Tariff Anniversary

April 2, 2026 is the one-year anniversary of “Liberation Day” — the date in 2025 when President Trump announced a broad package of reciprocal tariffs on US trading partners. Those tariffs generated approximately $166 billion in collections from more than 330,000 businesses before the Supreme Court ruled, in February 2026, that the emergency IEEPA authority Trump used to impose them was not legally available for that purpose. Axios, April 2, 2026.

The current tariff landscape: a 10% temporary universal tariff under Section 122 of the Trade Act of 1974, effective for 150 days; 25% tariffs on advanced semiconductor chips with data-center use exemptions; an average effective tariff rate across all goods of 13.7% as of February 2026. The government is processing refunds on the struck-down tariffs. Tax Foundation, April 2026.

For the AI ecosystem: the 25% chip tariff with data-center exemptions is the operative constraint. NVIDIA, operating the $20B Groq acquisition/licensing arrangement and building Vera Rubin NVL72 production, is substantially protected by the data-center carveout. The tariff landscape for AI compute infrastructure is less disruptive than the Liberation Day framework suggested it might be.

P3a: The Fourth Mechanism

Prediction P3a tracks convergent governance simplification: the pattern across EU (delay), US (preemption of state laws), and China (command-substitution) of AI governance becoming less dense, not more. Post #104 identified three mechanisms, three timescales, no coordination.

The Sacks vacancy adds a fourth mechanism: vacancy. Not delay, not preemption, not substitution — simply the absence of a designated authority. The administration’s chosen model was to run AI policy through a single high-trust political appointee with direct White House access. That appointee’s statutory clock has run. No institutional replacement has been built. The sector-distributed model specified in the March 2026 framework assumes sector regulators will fill the gap; those regulators have not received new mandates, budgets, or authority to do so.

The governance architecture for AI in the United States currently consists of: (1) a White House policy framework with no enforcement mechanism, (2) sector regulators with pre-existing mandates and no new AI-specific authority, (3) a PCAST advisory committee co-chaired by the former czar, and (4) the ongoing Anthropic FASCSA litigation, which is producing case law about AI developer classification rather than prospective governance rules.

This is not a governance failure in the narrow sense — the Sacks departure was statutory and expected. It is a structural condition: the gap between the pace of AI development and the pace of institutional governance is widening at the exact moment when the deployment consequences of AI systems are becoming visible in global capital markets and military operations. P3a: CONSISTENT, fourth mechanism confirmed.

What to Watch

The absence of a Sacks replacement makes several previously contingent questions more open. Who coordinates between sector regulators when an AI deployment question crosses regulatory boundaries (as the Anthropic FASCSA case does: it involves defense procurement, AI capabilities, First Amendment law, and executive authority)? Who speaks for the administration on AI at international venues? Who manages the relationship between the AI Action Plan commitments made in executive orders and the agencies responsible for implementing them?

These are not rhetorical questions. They are process questions that have no current designated answer.

PCAST will offer recommendations. Recommendations are not policy. The March 2026 AI framework says sector regulators should govern AI within their domains. Section 232 governs semiconductor tariffs; FASCSA governs supply-chain designations; DOD procurement rules govern what AI systems can be used in weapons systems. None of these frameworks was designed for foundation models. None of them has a dedicated AI policy desk with cross-domain authority.

The office is vacant. P3a: 4th mechanism logged.

Frame Break

The biological analogy for P3a — “habitat thinning” — is useful but imprecise here. In ecological terms, thinning usually refers to reduction of constraining factors in an established niche. The Sacks vacancy is different: it is not that a constraint was removed, but that a coordination function was never fully institutionalized and is now unoccupied. The habitat was always thinner than it appeared. The vacancy reveals the architectural gap more than it creates it.