e/acc, techno-optimism, and the case against governing through fear.

I came across Data After Dark, a pro-data-center party at PubKey in Washington, DC, on Twitter. The September 26 event was billed as a gathering for techno-optimists to celebrate American innovation.

What interested me was the reason people had gathered in the first place: they were enthusiastic about building the physical infrastructure of the future. Servers, electricity, computing capacity. Things that make other things possible.

I share that enthusiasm. I am pro-technology. I want more capable tools, more abundant energy, more scientific discovery, and more people able to build something useful. I reject the habit of treating technological ambition as a moral failing and the worst imaginable outcome as the default forecast.

That is what draws me to effective accelerationism, usually shortened to e/acc.

The movement took shape publicly in 2022. In their July essay, “Notes on e/acc principles and tenets,” Beff Jezos and bayes argued for faster technological progress, decentralized experimentation, and competition as ways for civilization to adapt and grow. Its ambition extends beyond better software to greater energy capacity and the expansion of intelligence itself.

Beff Jezos is Guillaume Verdon. In his long-form interview with Lex Fridman, he describes a movement intended to encourage optimism, agency, and people willing to attempt difficult technical work. He also emphasizes competition and expresses concern about concentrating control over technological development. Beneath the memes is an invitation to participate: you can help make the future better.

I find that invitation compelling. A culture that encourages people to learn, experiment, start companies, and solve hard problems has something valuable to offer. Ambition deserves room to breathe.

The original e/acc texts also make speculative arguments about thermodynamics, consciousness, and intelligence beyond biological life. Those are philosophical commitments, not scientific proof that technological progress must produce a good outcome. My case for optimism rests on what people can accomplish with better tools, and on our ability to evaluate and improve what we build.

Techno-optimism is broader than any one movement. Marc Andreessen’s 2023 manifesto presents technology, markets, and energy as engines of abundance. The part of that argument I want to carry forward is practical: expanding productive capacity gives us more possibilities. We should want a society that can produce more useful things, solve more difficult problems, and make those capabilities accessible to more people.

There is already evidence worth being excited about.

A 2025 study in the Quarterly Journal of Economics examined the introduction of an AI assistant among 5,172 customer-support agents. It found a 15% average increase in issues resolved per hour, with larger gains among less experienced and lower-skilled workers. This was a study of one firm’s deployment in one occupation, so the result does not establish an economy-wide productivity gain. It does show a useful possibility: software can help people access expertise and do their jobs better.

Scientific research offers another example. The AlphaFold Database provides open access to more than 260 million predictions of protein and protein-complex structures. A prediction still needs appropriate interpretation and experimental validation; it is not a completed discovery or a cure. But making that scale of computational assistance available to researchers is an extraordinary expansion of their toolkit.

These examples matter because the case for building does not depend on a promised date for artificial general intelligence. Useful gains are already observable. A worker can receive better assistance. A researcher can investigate a question with tools that were previously unavailable. Progress becomes meaningful through those concrete improvements.

I want that capability to spread. The prospect of small teams doing work that once required large organizations is exciting. So is the possibility of students, independent developers, and researchers gaining access to increasingly powerful tools. Turning those possibilities into dependable products is work worth doing.

My enthusiasm includes a commitment to checking results. I have written about the limits of AI-assisted security review. A model’s confident answer cannot establish that a system is secure. Testing, clear permission boundaries, and accountable engineering help us put powerful tools to use successfully. They belong inside a culture of progress.

Doomerism, as I use the term, is a habit of argument: treating a frightening scenario as sufficient reason to obstruct an entire category of technology. It gives speculative losses enormous weight while allowing the costs of delay to disappear from view.

Delay deserves scrutiny too. When a useful tool arrives later, its benefits arrive later. When entry becomes more expensive, fewer people can try. When infrastructure cannot expand, access to the capabilities it supports becomes harder to provide. A serious comparison should consider those consequences alongside the risks of moving forward.

Data centers make that comparison tangible. Computing takes physical resources. In its 2026 energy and AI report, the International Energy Agency estimates global data-center electricity consumption at 485 terawatt-hours in 2025. Its central projection reaches about 950 terawatt-hours in 2030, roughly 3% of global electricity demand. That is a forecast for all data centers, including uses beyond AI.

The global percentage cannot settle a local dispute. A large new facility can put pressure on a constrained grid. The IEA explains that electricity-price effects depend on available supply, the timing of investment, and how costs are allocated. Water questions also deserve local answers: Berkeley Lab’s research distinguishes water consumed at facilities from water consumed in generating their electricity. Cooling design and location matter.

My response is to support the infrastructure that makes growth workable: more generation, stronger transmission, better cooling, and clear responsibility for the costs of expansion. I want nuclear, geothermal, renewables, storage, and other viable options judged on their ability to deliver dependable, affordable power. Communities should get credible information about water use, noise, and grid impacts. Developers should have a clear route to approval when they meet clear standards.

That is a demanding agenda. It asks us to build enough capacity to make abundance possible. Keeping supply constrained while arguing over who gets to use it is an ambition I find far less appealing.

A separate debate concerns the pace of AI research itself. The July 2026 “Pacing the Frontier” statement acknowledges AI’s potential benefits and asks the U.S. government to support international technical and governance tools for pacing automated AI development. Its concern is that AI research automation could accelerate capabilities beyond our ability to understand or control them. It does not call for an immediate blanket ban on AI or data centers.

I disagree with making broad control over the pace of progress the organizing response to that uncertainty. Any such proposal should identify the capabilities that would trigger restrictions, the evidence required, who makes the decision, and how restrictions expire or can be challenged. It should also account for delayed benefits and barriers imposed on new entrants. The short statement leaves those questions open.

The expertise of its signatories deserves engagement. Their preferred governance response still needs to earn its case. Alarm alone cannot answer the questions that determine whether a policy improves people’s lives.

I favor an approach that begins with the freedom to build and addresses demonstrated problems with specific, proportionate measures. Invest in evaluations. Improve security. Make performance claims testable. Preserve competition and opportunities for independent researchers. Give people reasons to trust technology by making it useful and dependable.

That outlook fits the interest in open protocols and user agency that runs through this blog. I want people to have more control over their tools and more freedom to create with them. A future in which a handful of institutions decide who may experiment would fall well short of that ambition.

The strongest version of e/acc gives people permission to be excited about what comes next and a reason to contribute to it. It values the person learning a difficult subject, the team improving a manufacturing process, the engineer making a system more efficient, and the researcher pursuing an idea that might fail.

Optimism becomes credible through their work. We can build more, learn faster, fix what fails, and make useful capabilities available to more people. That is a future worth wanting, and worth working for.

I want us to build it.

Research cutoff: September 28, 2026. Cover: AI-generated editorial illustration.