.webp){kind=small}
Before I founded Axon Capital Management, my fiduciary advisory firm in Georgetown, I worked in corporate finance at Applied Materials, the semiconductor equipment maker. My job there was to help decide where capital should go. A billion-dollar investment arrived as a model: a stream of projected cash flows, a hurdle rate, a payback period, a list of risks that could push the schedule right. Nobody in those rooms was cavalier about money, but at that scale the numbers become abstractions, and a $10 million line item can disappear into a rounding assumption.
Today, through my work in financial planning, I sit on the other side of the table. The people I work with across Central Texas are deciding whether they can buy their first home, whether a second child means one parent leaves the workforce for three years, how to balance retirement contributions against a mortgage and a parent who needs help, and whether a fixed income will still cover property insurance in five years. In that world, $10 million is not a rounding error. A $15,000 gap is often what separates a family that builds wealth from one that slowly falls behind.
I have come to see capital investment from both sides of the ledger. The data center boom now spreading east of I-35 brings those two ledgers into closer contact than anything I have seen in my career. On the same stretch of farmland, companies are committing sums that dwarf the region's municipal budgets, while the families who live nearby are working through the arithmetic of rent, child care and insurance. My argument is that Central Texas has a rare chance to connect those ledgers, and that doing so is good business for everyone at the table.
The scale is easiest to grasp statewide. Economists at Rice University's Baker Institute estimate that annual data center investment in Texas rose from about $3.4 billion in 2022 to $52 billion in 2025 and roughly $66 billion in 2026, with a baseline projection near $100 billion a year by the end of the decade. The state comptroller's illustrative AI facility, a 400-megawatt building, costs about $11 billion.
Central Texas has claimed a growing share. A June survey counted 30 operating or planned data centers on nine campuses in Williamson County alone, eight of them operating at about 500 megawatts, with 1.6 to 1.9 gigawatts possible at build-out; in Bastrop County, roughly 5,500 acres around Cedar Creek are tied to data center plans, including EdgeConneX's campus and 1,229 acres Amazon bought in May. Status matters, and much of this is not yet real. EdgeConneX's first $440 million Cedar Creek building is essentially complete and leased to CoreWeave. Sabey, Switch and Skybox are operating or building in Round Rock and Hutto. Taylor's 664-acre Taylor Technology Campus, approved this month, will not begin construction until 2027 and expects first power in 2028. Many other projects are only filings.
The reasons are not mysterious. The Williamson County Economic Development Partnership attributes the cluster to the electrical backbone built for Samsung's Taylor fab: it is about the availability of power. Add large, inexpensive tracts, fiber, and proximity to an Austin technology economy that companies like Big Watt Digital describe as one of the country's most dynamic. Constraints are real, too. Gov. Greg Abbott ordered ERCOT and the Public Utility Commission this summer to audit queued data centers before new grid connections, and to require them to pay for their own infrastructure and reuse water. Hays County has paused incentives for high-water-use projects, and San Marcos has banned data centers outright.
This distinction between durable advantages and purchased ones is the foundation of everything that follows. Power, land and an existing technology ecosystem are hard to replicate. Local tax abatements are easy to match anywhere. Microsoft now pledges to pay full property taxes and seek no local abatements, and Taylor's developers asked for none. When the scarce input is something a community controls, the community has leverage, and it should use it thoughtfully.
Economic development has long been measured in two numbers: dollars invested and jobs created. Data centers break that scoreboard. EdgeConneX's first Bastrop building, a $440 million project, was projected to create about 60 permanent jobs, or more than $7 million of capital per job. Pacifico Energy's proposed $2.2 billion gas-and-data-center project in the same county listed 25. By contrast, SpaceX's planned $855 million Starlink expansion in Bastrop County is tied to at least 500 jobs. Even Samsung's Taylor fab, itself extraordinarily capital-intensive, projects about 1,800 direct jobs in its first decade.
None of this makes data centers bad neighbors. Their benefit arrives through taxable capital rather than payroll, and it can be substantial. Hutto officials estimate that a single data center generates property tax revenue equivalent to about 500 homes while requiring city services equivalent to about 15. Taylor's developers project $1.38 billion in city tax revenue over 20 years, about $69 million a year for a city that collects roughly $21 million today. That is a developer-commissioned forecast, not money in the bank, but the order of magnitude is plausible.
Two features of Texas law complicate the picture. Under the state's school finance system, new taxable property often reduces state aid or increases recapture rather than adding operating revenue for the local district. And the state forgoes sales tax on data center equipment and electricity, an exemption the Senate Finance Committee now projects will cost about $3.3 billion in the 2028–29 budget cycle. A better scorecard would ask three questions: what net revenue reaches local budgets after these offsets, what costs the project imposes on water, power and roads, and what measurable improvement residents see in their own finances. The third question is almost never asked.
The families I work with are not helpless. They are teachers and engineers, nurses and electricians, retirees who spent decades saving. They are working hard in a region that has grown faster than its systems for housing and caring for people. Their pressures differ by stage of life, and the data bear that out.
For young adults, the binding constraint is housing. About 52 percent of renters in the Austin metro spent more than 30 percent of their income on rent and utilities in 2023, according to Harvard's Joint Center for Housing Studies. Home prices have cooled, with a metro median of $412,000 in August and $399,900 in Williamson County, but cooler is not cheap. In my practice, the young professionals who cannot buy are rarely failing at budgeting. They are earning respectable salaries, carrying rent that crowds out saving, and watching a down payment recede faster than they can accumulate it.
For young families, child care is the shock. Travis County has the highest child care prices in Texas, with median infant center care near $12,000 a year; Williamson County was about $10,500 in 2022. Price is only half of it. When Travis County voters approved a child care tax in 2024, officials estimated almost 3,000 children under three lacked access and that the subsidized waitlist ran two years. A common conversation in my office is not whether a family can afford care but whether one parent's after-tax income, minus care, is worth the career it preserves. Too often the answer is made by a waitlist rather than by the family.
Essential workers sit in a squeeze of their own. An average two-bedroom apartment requires about $75,000 of income, while average teacher pay in three Austin-area districts ran from about $60,000 to $64,000. Round Rock ISD now partners with a foundation to subsidize staff rent.
Middle-aged and older homeowners face a different set of costs. Texas lawmakers raised the school homestead exemption to $140,000, and to $200,000 for homeowners 65 and older, in 2025, and long-tenured seniors also benefit from the school tax freeze. Insurance has moved the other way: the average Texas homeowners premium rose from under $2,000 in 2020 to more than $3,500 in 2026. In Georgetown, where about 28 percent of residents are 65 or older, the retirees I see are generally careful planners. What unsettles them is a fixed-income budget absorbing insurance, roof replacements and care costs that nobody modeled a decade ago.
Set these figures beside the capital flowing into the region and the contrast is stark. The gaps that determine household outcomes are measured in thousands or tens of thousands of dollars. The investments arriving next door are measured in billions.
When economic development agreements disappoint, the explanation is rarely as simple as bad faith. More often, cities and counties apply a negotiating framework designed for factories, warehouses and corporate offices to an industry whose economics are fundamentally different. Several recurring mistakes deserve attention.
Communities offer incentives without establishing whether they are necessary. Bastrop County granted EdgeConneX a 10-year, 75 percent property tax abatement, then later hired consultants to examine the fiscal implications of data center development. Meanwhile, another developer in the county, Pacifico Energy, withdrew its applications for county tax abatements without canceling its proposed project. Taylor's newest data center development proceeded without a reported property tax abatement. None of this proves that any particular project would have happened without incentives. It does suggest that local governments should demand compelling evidence that a concession is necessary before sacrificing future revenue.
They mistake announced investment for realized economic benefits. The Baker Institute's October 2026 research identified 749 announced data center facilities in Texas compared with 228 active ones. Not every proposed development will be completed, and even completed campuses may generate taxable values considerably below their headline investment figures. Much of their capital is invested in computing equipment that depreciates rapidly. School finance further complicates the benefit: additional local property taxes can reduce state education funding rather than translate into larger operating budgets for school districts. Economic development agreements should be evaluated using realistic operating dates, taxable values, and revenue projections rather than full-buildout announcements.
They sacrifice transparency or flexibility to secure an agreement. Northern Virginia's enormous Digital Gateway project had its rezoning approval thrown out after a court found that Prince William County failed to satisfy public notice requirements. In Texas, Taylor approved its October 2026 data center agreement shortly before considering new data center regulations, establishing project-specific rules in advance. Taylor also obtained enforceable environmental protections and community payments through annexation that it otherwise lacked the authority to impose. The example illustrates a genuine trade-off: negotiating certainty today can limit a city's ability to respond to new concerns tomorrow.
They fail to establish how promised benefits will reach residents. Additional tax revenue does not automatically translate into lower household tax bills or improvements in affordability. That depends on future budgeting decisions. Voluntary corporate donations may support worthwhile causes, but without defined commitments, independent oversight and measurable outcomes, communities cannot reliably evaluate their long-term financial benefit. At the opposite extreme, blanket restrictions can create legal and financial risks of their own. Hill County rescinded its data center moratorium in June 2026 after a developer filed a $100 million lawsuit challenging the county's authority.
These problems share a common feature: communities negotiate the immediate approval, incentive or investment announcement more carefully than the financial relationship that follows. Data center campuses can remain in place for decades, with changing tenants, equipment values, infrastructure demands and public costs. A successful agreement should anticipate those changes from the outset.
I am not proposing that data center companies fix housing or child care. They cannot, and it is not their job. I am proposing that communities treat the approvals and services they control as valuable, and negotiate for a modest, predictable, enforceable share of the value those approvals help create, directed at the affordability problems where money demonstrably works.
Taylor offers a starting point. With the site sitting in its extraterritorial jurisdiction, where the city has little authority, Taylor negotiated a voluntary annexation agreement instead of an incentive. The developers accepted binding standards on noise, light, setbacks and cooling with reclaimed water, plus $28.2 million in commitments: $13 million for city quality-of-life projects, $10.2 million for Taylor ISD programs, $1.5 million in scholarships, $1 million for career and technical education and internships, and smaller nonprofit grants. It is a real achievement for a city of Taylor's size. It also has limits. The money is largely fixed regardless of how large the campus grows, much of it is diffuse, and the agreement lets the developer adopt future rule changes that benefit it while the city cannot do the same. A stronger model would keep what Taylor got right and add four elements.
First, full taxation remains the foundation. For a project that would come anyway because the power and land are here, a tax abatement is simply a transfer to the developer. The first thing a community should ask for is that the project pay its full property taxes, and cities should publicly commit a defined share of the resulting new levy to lowering the tax rate or avoiding future increases. That requires no developer payment and no legal novelty, only council discipline. The cautionary tale is Prince William County, Virginia, where critics say promised rate reductions from data center revenue never fully materialized.
Second, contributions should scale with what is actually built. A flat sum negotiated before a shovel goes in either overcharges a small project or undercharges a gigawatt campus. A payment of, say, $15,000 per energized megawatt per year would cost a 100 MW building $1.5 million annually and a 1 GW campus $15 million, and it arrives only when the facility is operating and earning. At an assumed lease rate of $150 per kilowatt-month, that is roughly 0.8 percent of gross rent. A one-time payment at the first building permit can supplement it for capital projects that need money up front.
Third, the money should go into one professionally governed fund. Rather than dozens of developer-selected grants, payments should flow into a community prosperity fund held at a community foundation or a city housing finance corporation, with independent auditing, public reporting, and allocation priorities set by elected officials in public. Lancaster, Pennsylvania, executed the first public data center community benefits agreement of this kind, routing about $20 million through its county community foundation in installments before and after operation. Such a fund should concentrate on three priorities.
Part of each year's payment should build a permanent endowment. Sustaining 100 subsidized infant slots at about $12,500 each indefinitely requires roughly $28 million of principal at a 4.5 percent draw. Endowments are slow, but they outlive changes in ownership, and data center campuses change hands often.
Fourth, enforcement should be built in. Payments should be secured by escrow or letters of credit, tied to permit and energization milestones, and reported publicly each year. St. Louis attached financial clawbacks to its data center agreement in case economic projections fall short.
Texas law shapes how this can be done. Cities cannot condition a permit on payments unrelated to a project's impact; required infrastructure exactions must be roughly proportional under the Texas Supreme Court's Flower Mound decision and state statute. These must therefore be voluntary agreements exchanged for something the community legitimately provides: annexation, water and utility service, development agreements that offer regulatory certainty, or, where justified, a tax abatement. Counties, which cannot zone, have less to offer. The approach works best where Taylor's worked: in cities with services and approvals developers need.
I spent years in rooms where projects were approved or killed on the strength of their risk profile, and I would make the case to a data center board on financial grounds alone.
Start with the cost. Using the comptroller's benchmarks of $11 million to $27.5 million per megawatt, a 1 GW campus represents $11 billion to $27.5 billion of investment. Taylor's $28.2 million is about 0.1 to 0.25 percent of that. A $50 million package would be under half a percent. Announced capital is not profit, and much of it is tenant-owned IT equipment; in the comptroller's example, 64 percent of cost is IT. Even measured against the developer's smaller share of the capital stack, the cost stays near one percent.
Then compare it with the cost of delay. With record wholesale asking rents near $196 per kilowatt per month at the end of 2025 for smaller requirements; large hyperscale blocks lease for less. At an assumed $120 to $175, a 100 MW building that sits idle forgoes $12 million to $17.5 million of revenue every month, before the carrying cost of capital already spent. A six-month delay on one building costs more than Taylor's entire package two or three times over.
Delays are a serious risk. Data Center Watch counted 75 U.S. projects worth about $130 billion blocked or delayed by local opposition in the first quarter of 2026 and 45 worth about $68 billion in the second. Northern Virginia's 2,100-acre Digital Gateway died after courts voided a rushed rezoning, after the county had spent about $1.7 million defending it. Closer to home, Pacifico Energy withdrew its abatement requests in Bastrop County after more than 100 residents turned out against them, and opposition groups continue to press in Taylor.
A transparent, formula-based agreement does not buy a permit or excuse a project from air, water or grid standards; those must stand on their own, and no payment should substitute for them. What it can buy is predictability: a known price, a public process, and a constituency of residents who can point to child care seats and apprenticeships rather than only to trucks and transmission lines. For assets designed to operate for decades, that durability has real value. It is the same logic behind Microsoft's decision to pay full taxes and full power costs; its president has argued that buildouts progress only when communities believe the gains outweigh the costs.
The bargain should also be calibrated. A 25 MW enterprise facility like Blueprint's $160 million Georgetown project could not absorb the same payment as a gigawatt campus; $10 million would be more than 6 percent of its capital. Capacity-based payments solve this, and small facilities can reasonably be exempt. Some developers will still walk. Communities should let them, because a predictable framework applied evenly is worth more over time than any single deal.
The largest technology infrastructure investments of this generation are being placed in the ground now, on land that families in Taylor, Hutto, Round Rock and Cedar Creek drive past every day. These facilities will run for decades. The approvals that let them run are being granted in the next few years, and the leverage that comes with those approvals will not return once the concrete is poured.
Central Texas does not need to choose between welcoming this industry and taking care of its people. It needs a scorecard that counts both: full and fair taxation, infrastructure and environmental costs borne by those who create them, and a small, enforceable share of value invested where the evidence says it changes household outcomes — a child care seat that lets a parent keep a career, a home a teacher can afford near the school where she works, an apprenticeship that turns a high school graduate into a licensed electrician, a city budget that can hold tax rates steady for the retirees who built the place.
The servers in these buildings will train and run systems that may reshape the global economy, and much of the wealth they produce will be earned far from here. The power, the land, the water and the public trust that make them possible are local. If Central Texas is going to supply the foundation of the AI economy, the people who live on that foundation should own a piece of what it builds.
Article written by Brady Lochte, founder of Axon Capital Management and a fee-only fiduciary financial advisor serving individuals and families across the Austin metro area. Brady is committed to providing clear, transparent financial guidance that helps people navigate retirement, investing, and long-term planning with confidence.
This article reflects the independent research, analysis, and opinions of Brady Lochte, founder of Axon Capital Management, LLC, as of the publication date. It is intended for informational and educational purposes only and does not constitute investment, tax, legal, or other professional advice.
Financial figures, economic projections, and development estimates are based on publicly available information and sources believed to be reliable; however, their accuracy or completeness cannot be guaranteed. Announced investments, projected tax revenues, employment estimates, and proposed development agreements may differ materially from actual outcomes. Illustrative calculations and policy proposals are hypothetical and should not be interpreted as forecasts or representations of financial feasibility.
The views expressed are the author's own and do not represent the positions of any companies, government agencies, or other organizations discussed. References to specific companies or developments do not constitute investment recommendations or endorsements.
Please submit the form with any questions or comments and I'll be in touch shortly.
If you'd rather not wait, you can also book a time directly on my calendar using the link below:
Schedule a call
