V.E — Claims and Tests

The preceding sections developed an analytical apparatus. Part IV established the production function: how computation, energy, and data combine to produce cognitive output, and why Bitcoin mining creates a floor on electricity returns. Part V applied that apparatus to labor markets, settlement infrastructure, and physical constraints. The hurdle rate disciplines deployment. The V/C ratio orders automation. The actuation layer absorbs margin as cognition commoditizes. The term structure enables coordination at machine scale.

These are claims about structure, not descriptions of what has happened. Claims of this kind are useful only if they can be checked. The framework makes four predictions. Each implies observable consequences over the next decade. Each could be wrong.

---

The first prediction: a measurable floor exists beneath computational economics.

The Bitcoin network, through proof-of-work, establishes a continuously observable conversion rate between electricity and a globally liquid asset. Any computational workload generating less value per kilowatt-hour than mining—after accounting for hardware, cooling, and opportunity cost—is uneconomic at the margin. The floor varies with electricity price, network difficulty, and Bitcoin’s exchange value. But it exists, and it disciplines capital allocation in ways that capability metrics do not.

The test: observe allocation decisions at marginal facilities in power-abundant regions. If operators consistently run inference workloads generating negative returns relative to mining alternatives, and continue doing so over multiple difficulty adjustment cycles, the hurdle rate mechanism is not binding. Strategic deployments at flagship facilities, where other considerations dominate, do not falsify the claim. Marginal behavior does.

---

The second prediction: tasks automate in sequence of verification cost, not cognitive difficulty.

A task with high value and low verification cost—writing assessable by reading, code testable by execution, images evaluable by inspection—automates before a task with equivalent value but higher verification cost. The value-to-verification-cost ratio predicts the deployment frontier more reliably than capability benchmarks.

The implied sequence: machine-verifiable outputs (translation, summarization, structured data extraction) precede outputs requiring expensive human judgment or physical confirmation (diagnosis carrying malpractice exposure, advice carrying liability, assembly requiring safety certification). The automation of copywriting before surgery is not a failure of surgical AI; it is verification economics operating as expected.

The test: if high-verification-cost tasks routinely automate ahead of low-verification-cost tasks at comparable value, the ordering principle fails. Autonomous surgery becoming routine before autonomous copywriting would falsify the claim. So would autonomous construction inspection preceding autonomous document review.

---

The third prediction: rents migrate from the cognitive layer to the actuation layer as capability commoditizes.

When the cost of generating a decision approaches zero, binding constraints become: executing that decision in physical reality, proving correct execution, bearing liability for consequences, and obtaining permission to act. These bottlenecks do not commoditize at the same rate as cognition. Firms positioned at these chokepoints capture durable margin even as model providers compete toward commodity pricing.

The prediction implies observable market structure over the next decade: interconnection queue depth remaining binding through 2030, datacenter cap rates staying compressed below 6%, liability coverage for autonomous systems commanding premium pricing, and frontier lab gross margins compressing below 50% by 2028 while infrastructure operators maintain margins above 30%.

Evidence to the contrary would falsify the migration claim: abundant interconnection capacity clearing in under 18 months, infrastructure cap rates expanding above 7%, freely available autonomous-systems liability coverage at commodity pricing, sustained cognitive-layer margins exceeding actuation-layer margins through the decade.

---

The fourth prediction: autonomous agents require a common term structure to coordinate at scale.

Without a benchmark discount rate denominated in a neutral settlement asset, every bilateral agent interaction requires bespoke credit assessment. Coordination cost scales quadratically with agent count. A published yield curve for machine commerce—observable rates at benchmark tenors—collapses this to linear scaling.

The prediction: some form of Bitcoin-denominated term structure emerges at benchmark tenors (overnight, one-week, one-month, three-month, one-year) by 2030, and agentic commerce scales faster in ecosystems that adopt it than in ecosystems that do not.

Two conditions would falsify the claim. If the term structure fails to emerge and agentic commerce scales regardless, the coordination mechanism is unnecessary. If the term structure emerges and agentic commerce does not scale, other constraints dominate. Either outcome would require revising the framework’s account of what limits machine-to-machine coordination.

---

These four claims—the hurdle rate floor, the V/C ordering, the actuation migration, and the term structure necessity—form the load-bearing structure of the argument. Each generates predictions that can be checked against evidence as it accumulates.

The next part turns from mechanism to implication: where capital should position if the framework is correct, which assets appreciate, which business models survive the transition.