Bloomberg Reporter: Why “The Race Is Back On” Against China As U.S. Lifts Anthropic's Fable 5 International Restrictions

The Trump administration removed all foreign-access restrictions on Anthropic’s “Fable 5,” reversing a prior requirement for approval and lifting export controls for the model intended for public deployment. Earlier, access to the more powerful “Mythos 5” was only allowed for vetted entities (e.g., critical infrastructure and cyber defenders). The change is framed as a competitive reset—“the race is back on”—supporting potential broader enterprise integrations (e.g., Tenable partnerships) and helping Anthropic’s valuation ahead of a planned IPO, though security concerns remain about potential “distillation” by Chinese rivals.

Analysis

This reads less like a direct earnings change and more like a policy de-risking that lifts the terminal value of U.S. frontier AI distribution. The first-order beneficiary is the platform with the biggest incremental token demand and enterprise attach rate: broader overseas access improves utilization of the model layer, but the real monetization accrues to cloud and workflow owners that can wrap security, hosting, and deployment services around it. That is why AMZN matters more than the headline suggests; if international adoption rises, AWS inference, managed security, and sovereign-cloud deals can scale faster than pure model revenue.

The second-order winner is the AI security stack. If customers are now more willing to expose production workflows to frontier models, they also need monitoring, exposure assessment, policy enforcement, and audit trails, which is supportive for TENB and adjacent security vendors. The catch is that more distribution also increases the probability of commoditization via distillation, so the very policy easing that expands TAM may simultaneously shorten the duration of model differentiation. That argues for favoring picks-and-shovels over model names into any valuation rerate.

The contrarian point: the market may be overpricing the speed of conversion from access to revenue. Global availability does not solve procurement friction, data residency, or cost-per-query constraints, and those are the real gating items over the next 1-3 quarters. If usage metrics do not inflect or if overseas regulators push back, the valuation boost fades quickly; structurally, the bigger risk over 6-18 months is that easier access accelerates imitation and compresses margins across the AI stack.

Key Decisions for Investors

• Add AMZN on any 1-2% pullback, sized as a tactical 1-3 month long: the thesis is AWS inference and managed deployment monetization, not the model access story itself. Falsify if AWS capex commentary softens or AI services growth does not reaccelerate over the next two quarters.
• Initiate a small long in TENB for 3-6 months: wider frontier-model adoption should expand demand for AI governance and exposure-assessment tooling. Keep size modest because hyperscalers can bundle security features and cap upside.
• Consider a smaller 6-12 month speculative long in ICE only if compute-futures volumes start printing: the optionality is on financialization of AI capacity, but this is a second-order trade and should be sized below core longs.
• Avoid chasing any broad AI-beta basket on this headline; the cleaner expression is infrastructure/security over model adjacency. If the market rerates pure-play AI names without proof of overseas monetization, fade that strength.
• Set a watch item on Chinese response and distillation evidence: if competitive releases accelerate or security breaches emerge, the policy easing becomes a negative for valuation multiples across frontier AI rather than a positive.