Diagnosing Where the Project Can Fail

Passing an initial screen does not make an opportunity viable. It earns the opportunity further investigation. The next level of the decision language examines individual systems in greater detail through a series of diagnostic frameworks.

GTSQT, which stands for Grid, Tariff, Substation, Queue, and Timing, examines the power pathway. Electricity is not simply a commodity requirement. The relevant question is what physical, regulatory, contractual, and sequencing conditions stand between a project and energized capacity.

DLPT, which stands for Disclosure, Legitimacy, Permitting, and Transparency, examines the approval environment. Infrastructure projects increasingly depend not only on whether permits can technically be obtained, but whether the process itself is considered legitimate by governments, regulators, communities, and other stakeholders.

R³F, which stands for Route, Redundancy, Reputation, and Failure, examines connectivity. Fiber proximity alone does not establish network viability. Physical routes, network architecture, counterparties, redundancy, operational reputation, and the consequences of failure all matter.

VWPP, which stands for Vapor, Water, Plume, and Public, examines thermal and water considerations beyond the mechanical system itself. Water consumption, atmospheric effects, visible plumes, operating conditions, and public perception can influence the durability of cooling choices.

A⁴ focuses on alignments because infrastructure projects do not operate according to one schedule. Power, permitting, equipment, construction, financing, interconnection, customer commitments, and network delivery each follow different timelines. A project becomes executable only when enough of those schedules align at the right moment.

LENS²: Can It Work for This Client?

A project can be technically possible and still be wrong for a particular customer. LENS² examines Latency, Expansion Path, Near-Term Delivery, and Stability to determine whether a viable infrastructure opportunity actually works for the intended user.

Latency asks whether the location can serve the customer's network and computing requirements. Expansion Path considers whether the customer can grow if its requirements change. Near-Term Delivery focuses on whether meaningful capacity can actually be delivered when it is needed rather than at some undefined point in the future.

Stability asks whether the power, regulatory, political, infrastructure, and operating conditions are durable enough to support the intended commitment. Together, these considerations change the question from whether a site is generally good to whether it is appropriate for a particular customer, workload, scale, and moment.

Permission Has Become Infrastructure

The industry has historically been very good at thinking about physical infrastructure. It now has to become equally sophisticated about institutional infrastructure because communities are increasingly questioning electricity costs, water consumption, tax incentives, land use, noise, emissions, employment claims, transmission investment, and who ultimately benefits from development.

Those questions are not peripheral to the project. They increasingly determine whether the project happens. A development that cannot maintain political and institutional legitimacy may be just as constrained as one without a substation or adequate transmission.

That is why I describe permission as infrastructure. Approvals, public trust, transparency, and community acceptance cannot simply be addressed after the technical and financial strategy has already been determined. They increasingly belong inside the original development thesis.

The Local Balance Sheet

AI infrastructure can generate enormous investment, but national economic benefits do not automatically resolve local concerns. Communities experience infrastructure through a different balance sheet and reasonably ask what happens to electricity rates, who pays for grid upgrades, how much water is required, what tax revenue is created, how many permanent jobs remain, and what infrastructure the community receives in return.

They may also ask what alternative uses of land disappear, what environmental or operating impacts remain, and what happens if the market changes after major public or utility commitments have been made. These are not necessarily arguments against development. They are questions about how the costs and benefits of development are distributed.

Successful infrastructure development increasingly requires understanding both balance sheets, the one seen by investors and the one experienced by the community. Ignoring either creates risk that eventually appears somewhere else in the project.

The Missing Middle

Much of the industry's attention is concentrated at the extremes. Small edge facilities serve highly localized requirements, while enormous hyperscale campuses dominate headlines, capital flows, and discussions about power demand.

Between those extremes sits what I describe as the Missing Middle, particularly opportunities in roughly the 10 to 40 MW range. These projects can involve different customers, development strategies, existing assets, secondary markets, capital structures, and infrastructure pathways than the largest hyperscale developments.

As AI workloads evolve and inference becomes more distributed, the relationship between workload, latency, geography, power availability, existing infrastructure, and customer requirements deserves greater attention. Bigger will remain important, but bigger is not the only question the market needs to answer.

Toward a Common Decision Language

AI infrastructure brings together professions that do not naturally speak the same language. Developers, utilities, engineers, investors, operators, technology companies, lawyers, regulators, economic development agencies, elected officials, and communities each see a different part of the project.

The challenge is creating enough shared language that those parties can distinguish between an attractive narrative and an executable opportunity. That is the larger objective behind my work and the reason I have developed a decision language rather than another forecast about how many gigawatts the industry may eventually require.

In an infrastructure-constrained market, identifying demand is increasingly the easy part. Determining where that demand can become physical reality is the harder and more consequential question.