An Amazon Web Services data center is shown in Stone Ridge, Virginia. Virginia has aggressively courted AI investment. (Photo by Nathan Howard/Getty Images)
Getty Images
In 2025, artificial intelligence captured 61 percent of all global venture capital investment, pulling in $258 billion out of a $427 billion total market, according to a February 2026 report from the OECD. That’s double the share only three years earlier. AI companies that are based in the U.S. won an astounding 75 percent of that global AI venture capital. The EU finished second at 6 percent. So, it’s clear – U.S. states are ground-zero for the AI investment game.
The deals are big. Investments topping $100 million now make up about 73 percent of total AI investment value, showing a concentration in a handful of category leaders. Amazon, Google, Meta and Microsoft are expected to collectively spend roughly $700 billion on AI infrastructure in 2026 alone.
But scale of commitment is not the same as quality of commitment. Shockingly, a significant portion of that capital is being deployed with surprising imprecision. Goldman Sachs warned that many data center investments risk failing to generate expected returns if the industry can’t sufficiently monetize AI models. Oracle shares fell nearly 40 percent over a three-month period on fears that its massive AI data center bets won’t pay off. Remember the telecom overbuild of the early 2000s? It’s worth looking up.
Do we think these billions are being tossed around on whims, without adequate research and analysis? We do not. But: Is there a deficit of state-by-state, category-by-category data to be analyzed when deciding which state to invest in? Evidence suggests so.
Physical AI vs. Knowledge AI: Two Investment Theses, Dozens of Variables
A useful framework for AI investors is the distinction between Physical AI and Knowledge AI investment. Physical AI covers the hard infrastructure of the artificial intelligence economy: data centers, GPU farms, power substations, fiber networks, cooling systems, and the land and buildings that house them.
Knowledge AI focuses on the intellectual and human capital side — research and development, patents and IP, university ecosystems, STEM workforce pipelines and the startup and venture culture that turns ideas into companies.
On top of those, quality of life – cost of living, crime, healthcare availability and so on – should be considered. At least for now, humans are the essential piece of AI investment.
This all points to the exercise of state-picking: Which U.S. states should AI investors choose?
Both the Physical and Knowledge categories require rigorous location analysis, but they demand different inputs. Physical AI investors need to understand energy: a large data center can consume as much electricity as a city of 50,000 people, and power costs have risen nearly 30 percent since 2020. They need to evaluate water: hyperscale data centers are expected to consume between 16 billion and 33 billion gallons of water annually by 2028. They need to assess permitting speed — in some states, data center projects can move from approval to groundbreaking in months; in others, regulatory gridlock can stretch timelines by years. And they need to scrutinize the tax and incentive environment: at least 36 states now offer some form of data center tax incentive, but these packages vary enormously in scope and reliability. Texas provided more than $1 billion in data center subsidies in 2025.
Knowledge AI investors face a different matrix. They need to consider the density of research universities and the quantity and quality of STEM talent pipelines. They need to understand state R&D tax credits and technology transfer policies. They need to assess the health of regional startup ecosystems and the availability of technical workforce at scale.
What the Rankings Already Tell Us
Several industry sources have begun grappling with state-level AI competitiveness. CBRE’s North America Data Center Trends report for 2025 identified markets with low power costs — Atlanta, Charlotte-Raleigh, Dallas-Fort Worth, Austin and San Antonio — as “poised for accelerated supply growth.”
CNBC’s America’s Top States for Business rankings highlighted Iowa, Utah, Wyoming, Montana and Nebraska among the top energy-supply states AI data center. Site Selection magazine, in its 2025 sustainability and corporate competitiveness rankings, placed Texas and California at the top of state-level rankings.
These rankings are useful starting points. But they are, by necessity, narrow in their focus. A ranking of states by electricity costs alone tells you nothing about workforce availability. A ranking by permitting speed tells you nothing about water supply. What AI investors actually need is a comprehensive, multi-variable picture.
Seeing this gap, my practice, BGR Analytics, created a data product designed to give AI investors a comprehensive picture of the investment market. The index evaluates all 50 states plus the District of Columbia, drawing on data from more than 100 categories, pulling in states’ economic strength, business climate, workforce advantage, human capital, energy cost, quality of life and so on. We found that Texas topped the ranking, followed by Utah, Florida, California and North Carolina tied at fourth, and Virginia. I use AI tools to break out separate Top 10 rankings for Physical and Knowledge AI investment.
The days of picking an AI investment location based on a single variable, a site visit, or a governor’s pitch are over. With hundreds of billions of dollars flowing into the AI economy each year, and with the consequences of a misallocated billions growing, the only responsible approach is one grounded in data.
