You’re reading Superintelligent, a newsletter about AI. Click here to get the full edition in your inbox.
Anthony Bak is head of AI implementation at Palantir. Mehdi Alhassani is head of government affairs at Palantir.
The surest way to guarantee that artificial intelligence becomes a tool of the wealthy elite is to block the infrastructure that would make it cheap for everyone else. Across the country, though, a growing coalition of activists, local officials and leaders on both the left and right are fighting explicitly to halt data center construction, restrict energy development, stop microchip production and slow the build-out of the backbone upon which cheap AI will depend. Resolving this dissonance is key to ensuring the benefits and wealth of AI accrue to our entire country, not just an elite few.
The instincts of this coalition are understandable. AI infrastructure is local and visible, and it threatens to strain power grids, disrupt communities and raise concerns about environmental impact. This is why leaders like Sen. Bernie Sanders (I-Vermont) are calling for a moratorium on data center construction, and why Denver put calls like that into action. It’s why there is a bill under consideration in Washington state that would require data centers to bear the full cost of the grid upgrades needed to serve them. In the second half of 2025, 20 projects were delayed or canceled after pressure from community activists, and nearly 100 more are being attacked in courts. Twelve state legislatures are currently considering moratoriums on data center construction, including in New York and Virginia.
The coalition’s victories, though, would come at the expense of those it claims to protect. Blocking the construction of AI infrastructure guarantees a future where the technology is scarce and expensive. Expensive AI is exclusive AI. It is AI that Wall Street can afford but an electrician can’t. It is AI that consolidates wealth at the top, eliminates jobs at the bottom and never reaches the small businesses, tradespeople and rural communities that stand to gain the most from it. Retreating from technology will not solve the problem.
The only path to broadly shared prosperity from this technology runs through the unglamorous, politically thankless work of building out its supply chain: one that ends with businesses and individuals utilizing AI, but starts with mining, chip manufacturing, energy production, data center construction and a competitive ecosystem of large language models, including open source variants.
Competition is already driving down costs: When OpenAI launched GPT-4, prices were $60 per million output tokens. Today that price is $10. But that reduction is being financed by venture capital and massive expenditures on data centers that remain too scarce to meet demand. (The Washington Post has a content partnership with OpenAI.) The industry will need to sustain itself when the subsidy runs out; without sufficient infrastructure, AI companies will be forced to raise prices. Large corporations and institutions will absorb the cost. Everyone else — the small businesses, start-ups, nonprofits, schools and AI-curious experimenters — will be locked out. Or worse, they will be forced into an ad-driven monetization model, where the incentives drive the most insidious forms of social disarray. That is the true dystopian scenario. When AI is expensive, the organizations able to afford it will use it to consolidate their markets and push out smaller companies through greater efficiencies.
Exclusive, expensive AI will stifle small business; cheap AI will empower them to remain competitive. For the electrician spending his evenings on invoicing and permit applications, or the restaurant owner buried in scheduling and inventory, cheap AI could reclaim hours lost to administrative drag. The electrician would gain time and resources to hire a second truck; the restaurant might be able to stay open longer. When AI is cheap, small, rural hospitals will have access to the same cutting-edge tools used by larger, wealthier networks.
The concerns of local communities ought not be ignored. Indeed, the only workable solution is one that is genuinely responsive to them. Energy supplies should be ensured for local communities, and to make energy plentiful and cheaper for all, new nuclear power should be a focus along with more wind, solar and other sources. Beyond energy, other solutions could include tax programs that ensure the presence of data centers is a lasting benefit to surrounding neighborhoods, or equity structures that give the community a direct stake in new data centers. The best solutions will emerge closest to the people affected, and they are more likely to succeed than federal moratoriums or statewide bans that may answer political objectives but do little for communities.
History has seen this before. Railroad build-out in the 19th century made long-distance travel and freight shipment significantly cheaper. Electricity was at first a luxury only the wealthy could afford, but President Franklin D. Roosevelt’s 1936 Rural Electrification Act helped produce the largest middle-class expansion the world had ever seen. The internet build-out in the 1990s, supported by public and private investments, made internet connectivity significantly cheaper for Americans. AI is at the same kind of inflection point, and the clock is even more unforgiving. Those blocking this build-out are not acting in bad faith, but if they succeed, they will have built the very future they claim to fear.
The post Halting data center construction will entrench inequality appeared first on Washington Post.
