American Intelligence
The narrative that rapacious techno-capitalists are extracting benefits for themselves while dumping externalities on powerless rural communities against their will, is demonstrably false. The superintelligence ecosystem is motivated to gain the consent and good will of the communities in which they seek to operate. Over the last few years, hyperscalers and development groups have made substantial commitments to workforce development, public education, grid upgrades, water conservation, and other adhoc benefits across the country, on top of the standard (large) tax expenses they’ll incur.
For example, NorthPoint Development, the Missouri firm behind a proposed 1,300-acre campus in Hazle Township, Pennsylvania, has offered $10,000 to each of the roughly 4,500 households in the area. The cash payout sums to $45 million, on top of the $120 million in community payments over 15 years that start rolling in when the first building opens.
The industry is willing to fund local communities and local improvements but what those benefits should be, how they should be coordinated, and most importantly, how permanent they can become.
Smart States should move quickly to institutionalize these commitments.
One-time cash payments code as a bribe. Even though the offer is generous and provides direct, tangible benefits, it makes people ask, "what’s the catch here?" "What am I giving up?"
Sam Lyman of the Bitcoin Policy Institute, a former Treasury speechwriter under Secretary Bessent, has written a great policy paper on data center dividends. Sam shows that each year, a host county could send households a fixed share of the property tax it already collects from compute facilities in their locale. His research shows that every 1 gigawatt of compute (located in the average rural county) can support yearly transfers, or credits, worth thousands of dollars for each resident after general fiscal priorities have been covered.
I believe the Super Intelligence buildout can be shaped to do what Lyman is proposing, and more.
States have a political opportunity in front of them. They should make existing industry offers standard, permanent, and more generous for the American people they represent. In particular, there are two big opportunities: direct, permanent benefits in the form of sovereign wealth funds that continue paying out after construction, and dignified, long-term benefits in the form of industrial infrastructure that attracts job-heavy manufacturing back to host communities.
Alaska, North Dakota, Texas, and New Mexico all have endowments. Three of these states are reliably understood to be center-right states. The super-intelligence buildout provides the financial means to scale and create momentum for more states to create their own endowment structures fit for the moment.
Endowments seeded by the superintelligence buildout follow the same principle as resource royalties: data centers rely on scarce public resources like grid capacity, siting approvals, and water, making access analogous to subsurface mineral rights.
Data center project could contribute to two endowments: one belonging to the host county (receiving the larger share) and the other a statewide pool that aggregates a smaller share from every project that operates within the state, acting as a vehicle that benefits state residents.
The statewide pool performs two jobs. First, it guarantees a financial floor for counties whose projects underperform or experience some form of negative outcome, and second, it gives counties that don’t have their own compute campus but host transmission lines, substations, and other enabling infrastructure a small stake in the buildout.
This proposal would not require the imposition of any new taxes on American citizens. Lyman's key insight is to capitalize these funds with a fixed share of existing property tax revenue, so there is no additional cost to the companies. In Loudoun County, roughly $685 million in tax was collected on data center computer equipment in fiscal year 2024, alone. That's roughly $165 million in tax revenue per gigawatt of load. Since servers are replaced every three to five years, the new equipment will be taxed at or near full replacement value, so the base won't wear out as it would in a traditional factory. States without a tax on business equipment can use negotiated payments instead, or some other mechanism that suits their tax regime.
Each state should choose a structure that fits its own tax code and the preferences of its residents. Further, the funds should always be predictable and tied to the continued operation of the facility, so that they can grow over its lifecycle enabling them to become savings vehicles rather than development banks, with permanent principal invested in diversified portfolios to support direct benefits to state residents where appropriate.
States and counties should also seek to treat data center projects as the financing anchors for large multi-tenant industrial parks. They can size the shared infrastructure for the whole park from the start: power, substations, transmission, water, fiber, roads, and graded land. The hyperscaler's commitment would then pay for the share it uses. The State and local municipalities would help finance the extra capacity built to attract diverse, high-value, and job-heavy industrial tenants.
That extra capacity would be held open for industrial tenants for an agreed period. Where it could earn something in the meantime, it should. Excess generation could be sold into the wholesale market, which adds supply and helps lower prices in the region for everyone. If manufacturers want to claim the capacity, then the anchor has served its highest public purpose. If they do not, the hyperscaler would hold a right to expand into whatever remains unclaimed when the period ends.
It is unlikely that a hyperscaler would accept a secondary claim on a campus it helped pay for unless the terms take into account their fiduciary duties. To make the math pencil out for them, a hyperscaler should hold equity in the development entity in proportion to what it contributes, earning returns on the full infrastructure even though its own load will not use all of it. This arrangement would align incentives for the hyperscalers, the State, and residents. Because industrial tenants' payments generate returns on the hyperscaler's equity, the company gains when other industrial tenants arrive. If none do and the compute operator also declines to expand, the capacity is still available for sale on the wholesale market.
Power capacity, grid connections, water rights, fiber, and permitted land can support a battery plant, a food processor, or a semiconductor packaging facility as readily as an intelligence factory. Industrial tenants, though, will only build specific buildings and related equipment once they are financially committed.
Powered, permitted land is a formidable attractor. Competing with other states and countries with tax credits and other financial incentives is easy, but it is also easily reversible with changing political winds. A working industrial site with credible capacity is difficult to develop, incredibly scarce, and irreversible. This is the type of value prospective industrial tenants are actively seeking.
There’s another slightly speculative but increasingly very real opportunity to keep in mind when designing these industrial parks anchored by the superintelligence ecosystem. Advanced manufacturing will increasingly run on simulation, digital models of physical systems, AI-assisted design, and large-scale inspection data. For defense suppliers handling controlled technical data, that work is easier to protect when the data never leaves the site.Standard commercial data centers are not generally built to accomplish classified work, but an industrial park can be planned so that a secure computing enclave meeting federal requirements can be easily added onto it later. Parks designed that way would have something few places can offer. At a time when the federal government is highly motivated to reinvigorate the defense industrial base, keeping this in mind could be extremely lucrative for states and municipalities.
This anchor-tenant model mirrors post-WWII interstate natural gas pipelines, where long-term household demand financed baseline infrastructure while excess capacity drew in downstream manufacturers. The model behind it is simple: fixed costs are covered by an anchor tenant with long horizons and a good credit profile. The excess capacity attracts other users and the value created is far greater than what it was initially built for.
Gas-fired generators, fertilizer plants, glassmakers, and food processors set up shop along the pipelines. They paid for their gas and service but they didn’t pay the expensive upfront costs needed to build the infrastructure they benefited from.
The capital investment driving the AI buildout is an opportunity to implement the model intentionally, quickly, and at scale.
Capital investment driving AI buildout creates a once-in-a-generation opportunity for local communities. By shifting from ad hoc deals to permanent structures with state sovereign wealth arrangements and anchor-tenant industrial parks, state and local leaders can lock in lasting economic benefits for its residents, improve public infrastructure, and support the long-term reindustrialization of America.
