Voters overwhelmingly reject data centers, but big tech will likely still win
43 of 44 local referendums against data centers passed; even mayors and lawmakers were recalled. Yet capital, credit, and land advantages remain concentrated in Amazon and Google—the scarcer compute becomes, the better their position.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Even Texas, data centers' last refuge, is now blocking expansion
Texas was long viewed as data centers' last refuge: independent power grid, abundant natural gas, friendly policy environment. But the newly elected governor promised to pause new approvals until the state legislature reviews—and Texas's legislature meets only every two years. This could mean an indefinite halt. More alarming for incumbents: mayors and officials who backed data centers were recalled. Officials now fear losing their seats even without formal referendums, making them politically cautious regardless of policy.
— Ben ThompsonThe opposition isn't about facts; it's emotional and ideological
The root objection isn't factual. Like the ‘fake news’ phenomenon, voters don't believe false claims first and then oppose data centers; they harbor resentment toward tech elites first, then hunt for justifications. If one anti-data-center argument is debunked, they simply pivot to another—because what they really oppose is the entire AI-painted future being imposed on them.
— Ben ThompsonOnly people who love advertising will buy into AI's promises
The AI industry will only convince those already conditioned by advertising. Good advertising first convinces you that you have a need, then offers the solution; AI just asserts it replaces workers. Deeper problem: the industry treats human desires as completely modelable and substitutable variables—a planned-economy logic, not an understanding that humans perpetually generate new needs.
— Ben ThompsonData centers promise long-term gains, but big tech skips the short-term bills
Loudoun County exemplifies the data center virtuous cycle: local property taxes down, electricity down over a thirty-year arc. But the catch: data centers drive up local electricity costs and stress the grid in their early years. Big tech doesn't subsidize this transition—it demands communities subsidize them instead. For the industry's long-term story to be credible, it must first absorb the short- and medium-term pain it inflicts.
— Ben ThompsonSunk launch costs make space data centers suddenly rational
SpaceX's orbital data center concept mirrors shipping's marginal-cost logic: ship depreciation is sunk; if fuel and port fees are covered, sailing even at accounting loss makes sense. SpaceX has already built launch infrastructure for Starlink. Once that capital is treated as ‘already spent, not to be recalculated,’ sending server racks to orbit only requires covering marginal costs—suddenly the calculation shifts from absurd to plausible.
— Ben ThompsonCompute scarcity turns big tech's existing advantages into overwhelming dominance
In a compute-constrained world—where politics and grid capacity become limiting factors—incumbents dominate. Amazon and Google hold land, power, permits, and community relationships, enabling expansion without starting from zero. Their cash flows are thicker, credit ratings higher, borrowing costs lower. When raising equity, they have real profitable businesses backing the bet. By contrast, all the equity value of Anthropic and OpenAI rests on a single outcome: whether their models succeed.
— Ben ThompsonGoogle's pitch now echoes Meta's old playbook on scaling then monetizing
Google's latest earnings language mirrors Meta's old pitch on Stories and Reels: first perfect the product, scale it globally, then layer in advertising. Google Search's ads still live in traditional results pages; the AI mode carries almost no advertisements yet. But users query more frequently and spend longer in AI mode. Google's wager: once ads migrate into AI search, it becomes the new cash machine, as Reels did from initial drag to current engine.
— Ben ThompsonThe internet economy is shifting from mass scale toward intimate groups
AI makes building individual software easier, but because everyone can build, single vibe-coding apps lack scalable monetization. The real economic value flows to custom-built software for tight-knit groups. The hardest-hit businesses are those built on Google traffic for mass reach—Reddit recognized this shift and abandoned the pageview chase, pivoting instead to maximizing revenue per logged-in user. The internet economy is migrating from mass-scale reach to high-engagement small circles.
— Ben ThompsonIn their own words · checked verbatim
If you love advertising, you're optimistic about AI.
Ben Thompson9:19
call me a big tech shell, a data center is great for a community. That's the thing. And like, these are basically the cleanest and most profitable industrial projects in American history.
Ben Thompson20:51
If you want to get a data center built, put money in people's pockets.
Ben Thompson29:51
Starlink and SpaceX is the ultimate hammer looking for a nail company.
Ben Thompson36:27
I think Google is transforming as a company, like right in front of our eyes.
Ben Thompson1:08:30
the part of the company that seems most threatened is somehow the most capable of delivering day after day month after month
Ben Thompson1:10:30
AI is like a machine gun. You can train on your weaknesses.
Ben Thompson1:35:54
Figures
| Local anti-data-center referendums that passed | 43 of 44 | 1:05 |
| Texas projected share of U.S. data center capacity by 2030 | 25% | 4:13 |
| Google Malaysia data center investment | $15 billion | 24:57 |
| Amazon capital expenditure vs. free cash flow | $200 billion vs. $180 billion | 44:49 |
| Alphabet market cap loss after earnings | Over $500 billion; stock down over 13% | 59:18 |
| Stripe transaction volume as share of global GDP | 1.7% | 37:28 |
| ChatGPT search queries relative to Google | 20% | 1:13:34 |
| Reddit stock decline after earnings | ~30% | 1:14:34 |
Glossary
- smiling curve
- Internet market structure where value concentrates at both ends—mega-platforms and niche communities—while generic middleware gets squeezed.
- Neoclouds
- Emerging cloud startups renting compute capacity specifically to AI model builders, outside the incumbent giants like AWS or Google Cloud.
- Google Capital Company
- Google's strategy of channeling search profits into capital investments across cloud, semiconductors, space launch, and AI startups.
- group chat era
- An era where high-trust small groups become the primary organizing units for social and software use, displacing mass platforms.
How to listen
Investors tracking AI infrastructure policy and cloud vendor competition, and practitioners working on big tech strategy or government relations.
1:15:37-1:18:38 — a digression on X's rebrand from Twitter and corporate naming patterns—can be skipped, as it doesn't connect to the main data center and Google narrative.