Part VI · The Investment Thesis
VI.B — Positioning by Capital Type
9 min read · 1,719 words
The framework from VI.A identifies where structural advantage resides: not in cognitive capability, which commoditizes, but in the complementary assets that cognitive capability requires and cannot replicate. The question is how to position capital given that structure.
The answer depends on understanding what erodes, what appreciates, and where the gap between them creates opportunity.
Begin with what erodes.
Cognitive capability follows a gradient from raw capability to embedded workflow. At one end: frontier general-purpose models competing on benchmarks, where each improvement diffuses and the window between breakthrough and commodity compresses. At the other end: capability embedded in enterprise operations with proprietary data, regulatory approvals, and switching costs, where the cognitive layer may be commodity but the surrounding assets are not.
The gradient creates a discriminator: can the position retain customers if a competitor offers materially better capability at the same price? If yes, the position has accumulated complementary assets. If no, the position rents its revenue from the current capability frontier and faces erosion as that frontier advances.
Three categories follow. Frontier capability positions represent the highest-variance bets—a new architecture or training method could create temporary advantage, but the window between breakthrough and diffusion is compressing, and entering at frontier valuations risks exiting at commodity valuations. Interface wrapper positions take an API, add a user interface, and sell to a vertical; they face margin compression from both directions as the provider competes or raises prices and cheaper alternatives emerge. Full-stack positions combine cognitive capability with distribution, data, verification, and institutional relationships; enterprise contracts with switching costs, trained deployment teams, regulatory approvals, liability coverage—the complementary assets that capture returns as the core technology diffuses.
The statelessness of agents explains why cognitive positions erode so rapidly.
A common error in AI investment analysis is treating the agent itself as the defensible asset. This error persists because agents exhibit remarkable capability, and capability improvements seem continuous with prior technology waves where capability conferred advantage. But the agent-as-configuration has no memory of its prior invocations, no accumulated learning from deployment, no relationship with users that persists beyond what is externally stored. The “same” agent instantiated by a competitor with access to equivalent models, prompts, and tool bindings is functionally identical. There is nothing to defend because there is nothing that persists to be defended.
The moat, if it exists, lies in what surrounds the agent rather than in the agent itself: the data that feeds context windows, the workflow integrations that determine when agents are invoked, the institutional relationships that grant access to proprietary information, the user trust that permits agents to act with real consequences. These are the complementary assets. They exist outside the agent, in the infrastructure and relationships that the agent requires but cannot itself create.
A healthcare company deploying diagnostic agents has a moat not in the agents themselves but in the patient data that informs their context, the clinical workflows that trigger their invocation, the physician relationships that interpret their output, the liability coverage that permits their deployment, and the regulatory approvals that authorize their use. A competitor with identical models faces years of infrastructure building before matching these positions. The agent is commodity; the permission stack is not.
This analysis inverts the common venture framing. The question is not “which agent is most capable” but “which principal controls the context, integration, and permission infrastructure that capable agents require.” Capability without context is a demonstration; capability with context is a business. The demonstration can be replicated; the context compounds.
Verification cost predicts which domains commoditize first. High value-to-verification-cost tasks automate early; the cognitive layer commoditizes quickly; position in distribution and integration. Low V/C tasks automate slowly; the actuation bottleneck binds tightly; position in verification, liability, and physical infrastructure.
What appreciates is the inverse.
As cognition commoditizes, value migrates to what cognition requires but cannot replicate. Physical throughput—energy, fabrication, construction, logistics—cannot be instantiated by software. Lead times bound deployment regardless of how fast capability improves. Trusted interfaces—settlement rails, authorization chains, regulatory licenses—gate participation in economic activity. Verification infrastructure—sensors, attestation, audit mechanisms—closes the loop between cognitive output and physical reality. Liability capacity—insurance, bonding, legal structures that absorb risk—enables deployment in high-stakes domains.
The bottleneck premium accrues to whoever controls these assets. The premium is not uniform across capital structures. Infrastructure funds can access physical throughput directly. Venture capital can access early-stage verification and identity infrastructure but requires longer time horizons and different expertise than software-focused partnerships typically possess. Public market investors can access established utilities and manufacturers trading at lower multiples than their position in the production function may warrant. Growth equity can access scaling bottleneck owners—a robotics company with regulatory approval in limited jurisdictions needing capital to expand, a sensor network with coverage in one geography needing capital to build density.
The cognitive layer is where excitement concentrates. The actuation layer is where returns compound. The mismatch between where capital flows during installation and where value accrues during deployment creates opportunity for those positioned to capture the shift.
The spread between what erodes and what appreciates is calculable.
The hurdle rate from IV.E establishes a floor. A kilowatt-hour routed to inference must generate more value than the same kilowatt-hour routed to Bitcoin mining, or the capacity routes to mining. Cognitive layer investments face continuous margin pressure as this floor disciplines pricing. Applications that cannot maintain spread above the floor become uneconomic regardless of their technical capability.
Actuation layer investments face different economics. Physical throughput, regulatory licenses, and liability capacity do not compete directly with mining for electrical capacity. Their returns are governed by scarcity of the complementary asset. But the hurdle rate disciplines indirectly: actuation investments are valuable because they enable cognitive deployments that clear the floor. If no cognitive deployments clear the floor in a given domain, the actuation assets in that domain have no customers.
The spread between actuation margins and cognitive margins determines which vertical integrations are profitable. As cognitive costs fall while actuation costs remain sticky, the spread widens. Vertical integrations that combine commodity cognition with scarce actuation capture the spread. Geographic concentration follows: the locations where power is cheapest and most reliable are the locations where cognitive deployments most easily clear the floor; infrastructure in those locations captures the widening spread.
Within this structure, one opportunity stands out.
Multi-period contracts require a benchmark rate for discounting future obligations. Absent a shared rate, N agents must negotiate bilateral credit curves—the O(N²) explosion that stalls market formation. A common benchmark collapses this to O(N). The benchmark becomes infrastructure that every participant requires but no participant can unilaterally provide.
The Bitcoin yield curve, if it emerges, occupies this position. The firms that establish and publish the benchmark rate will capture structural advantage analogous to the creators of dominant reference rates in traditional finance. Liquidity concentrates around the reference. Early positions in this infrastructure compound as adoption scales; late entry means competing against established network effects.
The opportunity has distinct layers. Rate-curve construction requires observable market prices for Bitcoin-denominated obligations at multiple tenors. The firms that make these markets, publish the rates, provide the data feeds and calculation methodologies—they occupy gatekeeper positions. Collateral management for overcollateralized bonding requires custody, monitoring, and liquidation infrastructure; recurring revenue from every bonded transaction. Oracle infrastructure that connects on-chain contracts to off-chain reality captures value from every contract requiring external attestation.
This opportunity is largely invisible to traditional venture capital because it does not look like a software company. It looks like financial infrastructure: exchanges, custodians, data providers, calculation agents. These require different expertise, different fund structures, and different time horizons than software-focused portfolios accommodate. The opportunity is more accessible to growth equity backing scaling financial infrastructure, to public market investors positioning in established exchanges and custodians, and to infrastructure funds financing the physical and institutional substrate.
The window for entry is during emergence, not after consolidation. Once the benchmark establishes, the infrastructure positions become difficult to displace.
Time horizon determines which of these positions are accessible.
Venture capital with standard fund structures—ten-year life, three-to-five-year deployment, pressure for early markups and exits—has limited access to positions that require longer gestation. The fund that backs a robotics company with hardware development and regulatory approval timelines needs patience that standard structures do not provide. Growth equity with three-to-seven-year horizons can accommodate bottleneck positions already generating revenue but faces liquidity pressure if the transition takes longer than expected. Public market investors with permanent capital can hold positions through extended underperformance while waiting for structural dynamics to assert themselves—if they have conviction about the structure. Infrastructure funds with ten-year or longer horizons have structural alignment with physical assets that appreciate slowly but durably.
The value of patience increases as the transition moves from installation to deployment. Positions that appear to underperform during the installation period—when capital floods toward the exciting rather than the durable—may outperform during deployment as the structural dynamics assert themselves. The discount rate at which future bottleneck value is priced relative to current cognitive excitement creates opportunity for capital structures that can access it.
The framework identifies where structural advantage resides and which capital structures can access it. Venture structures can access early-stage bottleneck infrastructure but face time-horizon constraints. Infrastructure structures can access physical throughput with appropriate patience but cannot access early-stage opportunity. Public market structures can access established bottleneck owners but face timing risk relative to momentum.
The structure does not predict which firms will win. It identifies which positions compound and which erode. The logic is simple: cognitive capability commoditizes; the complementary assets appreciate; the term structure, if it emerges, creates infrastructure that captures returns from every participant. Capital that can access these positions before the transition from installation to deployment captures the spread. Capital that cannot waits for the opportunity to be arbitraged away.
The misalignment between standard capital structures and the positions the framework identifies as valuable is where opportunity concentrates. The question for any investor is which positions their structure allows them to occupy—and whether they have conviction about the structure itself.