Article

Build the Damn Data Centers: Progress Is the Point

America cannot lead the AI era while refusing to build the power and compute infrastructure it requires.

10 min readWatch the video →

Holden's opinion

Build the damn data centers.

America cannot demand the benefits of artificial intelligence and then panic over the physical machinery required to produce them. AI needs compute. Compute needs chips, power, cooling, networks, buildings, and skilled people. That is infrastructure. The countries that build it will own more of the capability, learn faster, and shape what comes next. The countries that refuse will become customers of the ones that did not.

I am not interested in playing the middle here. Progress is disruptive. Some jobs change or disappear. Some companies fail. New work and new industries appear. That has happened through every major wave of mechanization, electrification, computing, and automation. The answer is to adapt and keep building, not to freeze the economy because the transition makes people nervous.

China is not slowing down for us. It is expanding compute and power capacity as strategic infrastructure. If America talks itself out of doing the same, that is not caution. It is surrender.

Progress has always needed machinery

Every technological revolution arrives with a physical buildout. Electrification needed power plants, lines, motors, and redesigned factories. The digital economy needed fiber, server rooms, undersea cables, towers, and data centers. AI is no different. The software may feel weightless when it appears on a screen, but the work happens on machines in buildings connected to enormous power and networking systems.

The fear pattern is familiar too. New machinery threatens existing tasks, business models, and skills. That disruption is real. It is also incomplete as a story about what happens next. Research published in The Quarterly Journal of Economics found that the majority of current U.S. employment is in job specialties introduced since 1940. The Bureau of Labor Statistics' review of technological change describes displacement, productivity gains, and the creation of new tasks as forces that operate together. Technology changes the composition of work; it does not preserve an economy in amber. (QJE; BLS)

If the standard is that nobody can be disrupted, no serious infrastructure or productivity improvement will ever pass. That is not protection. It is a decision to let other places move first.

China is building while America argues

China's State Council Information Office reported that the country's intelligent computing capacity reached 2,185 EFLOPS at the end of June 2026, up 177% from a year earlier. The International Energy Agency projects China's data-center electricity use to rise by around 175 terawatt-hours from 2024 to 2030, a 170% increase. China and the United States together account for nearly 80% of the projected global growth. (SCIO; IEA)

Those numbers are not interesting because bigger is automatically better. They are interesting because they reveal intent. China's 2026 action plan explicitly coordinates reliable energy supply, clean power, and computing facilities. It understands that AI leadership is not only a model benchmark. It is the ability to run those models at scale and put them to work throughout an economy. (China National Data Administration)

America has the talent, capital, companies, and demand. What it cannot afford is self-imposed paralysis.

The propaganda playbook is already here

OpenAI reported in June 2026 that it had banned accounts likely originating from China after they supported an apparent covert influence operation aimed at American AI debates. One cluster, which OpenAI named Data Center Bandwagon, generated posts and images claiming AI data-center construction was raising electricity prices for ordinary families. The operation posed as Americans while hiding its origin and motivation. OpenAI said it found no evidence that the campaign achieved meaningful breakout, but the attempt itself is documented. (OpenAI)

My opinion is that most of the anti-data-center panic circulating online is propaganda, recycled propaganda, or people doing the propagandist's work for free. The documented campaign proves the playbook is real: borrow a legitimate local concern, remove the engineering context, make the infrastructure itself sound illegitimate, and encourage delay.

Anyone helping stop American compute capacity is helping China, deliberately or not. That is the outcome that matters.

Water is an engineering problem, not an argument against progress

A line like “data centers use a bunch of water” sounds decisive only when every important detail is removed. Facilities use different cooling systems, water sources, climates, and power supplies. Designs change.

Microsoft announced a closed-loop chip-cooling design that consumes zero water for cooling. The company says the design can avoid more than 125 million liters of water per facility each year compared with its fiscal-year 2024 fleet average. That does not describe every facility operating today. It demonstrates that water use is not fixed and that engineering can materially change the footprint. (Microsoft)

The answer is straightforward: choose sites that can support the project, build or contract the power, select the right cooling architecture, reuse water where it makes sense, measure the actual impact, and make the operator carry the incremental cost. Then build.

The choice is build or depend

Data centers will be built because demand for compute is growing and the capability is too valuable to ignore. The real choice is where they are built, who controls them, and which economies compound the learning and productivity they enable.

There is no version of technological leadership that avoids infrastructure. There is no version of progress that avoids change. America should stop apologizing for wanting to build the future.

Build the power. Build the grid. Build the compute. Build the damn data centers.

Sources

Full episode transcript

Build the damn data centers. I am tired of watching this country demand the future and then panic every time somebody tries to build the machinery that makes the future possible. AI needs compute. Compute needs buildings, chips, power, cooling, and people who know how to run it. That is not a scandal. That is infrastructure. China understands that. They are building. We are holding another meeting about whether progress feels scary. My position is simple: build them, build them fast, and stop acting like opting out is a serious strategy. People do this at every technological revolution. The new machine shows up, the old way of working gets threatened, and somebody declares that this time progress is going to destroy the economy. Some jobs do disappear. Some companies fail. People have to learn new things. That is disruption, and disruption is not pleasant for everybody. But the answer has never been to freeze the economy at the old job titles. The places that build, adapt, and reorganize capture the new productivity. The places that refuse get to buy the future from somebody else. The historical pattern is not that technology leaves every job untouched. Of course it does not. It destroys tasks, creates tasks, changes industries, and forces people and companies to move. American labor research shows that new work keeps appearing as technology changes what the economy can do. That is the part the doom story always leaves out. If your standard for progress is that nobody can ever be disrupted, then you have outlawed progress. I do not accept that standard, and neither should a country that expects to lead. Now look at China. China's own government says its intelligent computing capacity reached two thousand one hundred eighty-five EFLOPS at the end of June, up one hundred seventy-seven percent in a year. You do not have to admire their system to understand what that number means. They are treating compute capacity like strategic infrastructure. They are coordinating data centers with power, networks, and industrial policy. They are not asking whether the AI era should have buildings. They are deciding how many they can get online. That is the competition. China builds generation, transmission, compute clusters, and the connections between them. America has the capital, the companies, the talent, and the demand — and then lets every local panic turn into a national strategy. We cannot regulate ourselves into technological leadership. We cannot litigate our way into more compute. And we definitely cannot win by telling the companies building the future to put the machines somewhere else. Somebody will build this capacity. The only question is whether we intend to own enough of it. And yes, I think most of the anti-data-center panic people are swallowing is propaganda, recycled propaganda, or people doing the propagandist's job for free. We now have a documented example. OpenAI caught a PRC-origin operation posing as Americans and pushing claims that AI data centers were raising ordinary families' electricity prices. OpenAI called it Data Center Bandwagon. At the same time, China is expanding its own compute and power infrastructure. My conclusion is simple: anyone helping slow America's buildout is helping China, deliberately or not. The message works because it borrows real concerns and turns every one of them into a veto. Electricity prices. Water. Jobs. Land. Noise. There is always a reason the next piece of infrastructure must be built anywhere except here. Of course building something has an impact. That is what building means. The grown-up response is to engineer the impact, contract around it, and make the operator carry the cost it creates. The unserious response is to pretend that a country can have unlimited digital capability with no power plants, no wires, no cooling, and no buildings. Take water. People say data centers use a bunch of water as if every facility has the same design forever. It is nonsense. Microsoft has already designed a new generation of data centers that consumes zero water for cooling by circulating water in a closed loop. That does not make water irrelevant. It proves the footprint is an engineering decision, not a law of nature. Site the project correctly. Use the right cooling system. Reuse water where it makes sense. Then build it. A solvable engineering problem is not a moral argument against progress. This is not complicated. If a data center needs power, build or contract the power. If it needs grid upgrades, put the cost where it belongs. If the local water system cannot support one design, use a different design or choose a different site. Measure what the facility actually uses. Enforce the agreement. Then move. Those are construction and operating decisions. They are not a reason to spend five years debating whether America is allowed to possess the infrastructure behind its own technology companies. And we are not building these things for a toy. We are building a general-purpose capability that will help people design, decide, analyze, automate, and discover. We still underestimate what these models are going to be able to do because the capability is moving faster than most organizations can absorb it. That is an argument for more capacity, not less. The countries with compute will experiment more, deploy more, learn faster, and create things the countries without it will have to purchase or beg to access. There is no safe third lane where America refuses to build data centers, avoids every disruption, keeps every old job exactly as it is, and somehow remains the technological leader. That option does not exist. We can build the infrastructure and shape what happens next, or we can slow ourselves down and become dependent on the countries that did not. I would rather deal with the problems created by moving forward than inherit the much larger problem created by falling behind. Every generation thinks its disruption is uniquely terrifying. We are not special. We will be fine — but only if we keep moving. Some jobs will change. Some old ways of doing business will die. New work, new companies, and new abilities will take their place. That is what progress looks like while you are living through it. China is not going to stop because Americans found a scary water graphic online. Neither should we. Build the power. Build the grid. Build the compute. Build the damn data centers. Progress is the point.