The Download Covers Anthropic’s IPO, Smarter Power Grids, and Exercise Pills

Meta description: Anthropic filed a 335-page IPO prospectus highlighting risks, compute costs, revenues, losses, and voting control amid AI, energy, and science news.

Tags: Anthropic, IPO, AI, solid-state transformers, MIT Technology Review

Featured image: abstract circuit-like grid blending transformer silhouettes, AI nodes, and solar panels in deep blue and amber.

The Anthropic IPO filing came after weeks of speculation, when Anthropic submitted a 335-page prospectus, and The Download—its weekday newsletter delivering a daily dose of what is happening in technology—used the filing to ask whether the company’s story can survive public-market scrutiny. The Anthropic story ran in The Algorithm, the weekly AI newsletter that readers can sign up to receive in their inbox every Monday. Will Douglas Heaven framed the filing as a test of three competing images: an AI safety organization, a profit-driven megacorp, and a group hell-bent on creating superhuman technology.

Background: Years of Competing Anthropic Narratives

The prospectus, with about 100 pages focused on risks, lays out Anthropic’s view of the technology it is building and how it believes that technology will change the world. For years, the company has refined a public message that acts as a master class in narrative capture, presenting different faces to different audiences. Those faces carry huge tension: safety work, commercial scale, and a drive toward superhuman systems. The question is whether the IPO will reveal the company behind the mask or lean deeper into myth-making.

Development: Prospectus Risks, Losses, and Control

CNBC said Anthropic’s blockbuster IPO listing could determine the future of the AI investment boom. BBC reported that the filing warns of catastrophic AI risks and huge compute costs. Reuters reported that it exposes safety failures and a month-long training pause. The New York Times said the prospectus reveals soaring revenues but also staggering losses. Verge reported that Anthropic’s founders will retain voting control after the IPO.

Kate Brennan, who leads policy at the AI Now Institute, told the New York Times that the voluntary safety pledges AI executives recently made are toothless and unenforceable: “It’s as if they’re making a promise and crossing their fingers behind their backs.”

Latest Progress: Grid, Medicine, and Science Signals

Casey Crownhart’s story on solid-state transformers explained why they matter for the AI boom. Transformers are a crucial but nearly invisible part of the United States grid, helping efficiently move power across long distances. They are complicated and in short supply, and their basic design has changed little in about a century. As electricity demand rises from electric vehicles, data centers, and more, researchers and companies are looking to modernize these grid staples. Solid-state versions could ease supply concerns and cut power losses. The technology is still in development, but it could become a crucial part of the future grid.

Cassandra Willyard examined the idea of exercise in a pill. Exercise builds muscle, strengthens bones, lowers blood pressure, and boosts mood. By deciphering how movement benefits the body, researchers hope to develop medicines that offer some of the same perks without the workout. The idea is not to replace exercise but to bring its upsides to people who struggle with frailty, disability, or chronic illness. That story is part of “Are we there yet?” in the new Impossible issue of MIT Technology Review, a series that takes stock of six technologies that have long promised to change the world and scores them on a progress meter. Over the next six days, The Download will explore each one.

The must-reads were picked as the day’s most fun, important, scary and fascinating technology stories. Solar is quietly overtaking nuclear as a global power source, with cheap panels and batteries accelerating the shift, BBC reported. MIT Technology Review said the balcony solar boom is coming to the United States. OpenAI announced three new science discoveries that span mathematics, computer science, and physics, Verge reported, and Gizmodo said it enlisted expert “shepherds” to publish its breakthroughs. Bloomberg reported that Microsoft is offering buyouts to a quarter of staff amid rising AI costs, with around 60,000 employees eligible, while the New York Times said Microsoft plans $190 billion in AI spending this year.

The Los Angeles Times reported that company documents show the plastics industry knew microplastics could enter the body decades ago, with concerns dating back to 1974. MIT Technology Review noted that no part of the globe is free of human fingerprints. The Guardian reported that a demonstration of lighting Earth from space sparked alarm, with startup Reflect Orbital using space mirrors for the California test. MIT Technology Review said the tech could jeopardize the night sky. Associated Press reported that the United States designated helium a critical material because of its importance to semiconductor production. Ars Technica said hundreds of Boston Dynamics staff protested a deal involving U.S. Immigration and Customs Enforcement, calling for a halt to sales of the robot dogs. The Economist said treatments for age-related infertility are being developed at last, with two startups aiming to reduce chromosome errors in eggs, and MIT Technology Review explored how tech transformed babymaking. Wired reported that Grok says its AI-generated Odyssey is “not historically accurate,” with the bot itself admitting that Elon Musk’s filmmaking plan has failed.

What to Watch: Safety, Energy, and Human-Evolution Doubts

The filing’s risk section, the quote on voluntary pledges, and the grid story point to the same constraint: AI is becoming a physical, financial and safety problem at once. Solid-state transformers, helium, solar, batteries and compute costs show that model progress depends on power, materials and manufacturing. Anthropic’s IPO will test whether its public story can absorb that complexity without losing investor confidence.

Ben Crair’s “One more thing,” introduced under the label MCKIBILLO, turned to the problem with thinking you are part Neanderthal. The theory that many people have an “inner Neanderthal” rests on Homo sapiens and its cousin species once breeding, leaving some people today with a trace of Neanderthal DNA. That DNA is arguably the 21st century’s most celebrated discovery in human evolution. But in 2024, a pair of French geneticists called the theory’s foundations into question, proposing that what scientists interpret as interbreeding could instead be explained by population structure—the way genes concentrate in smaller, isolated groups.

The newsletter’s lighter corner, “We can still have nice things,” offered comfort, fun, and distraction to brighten the day and invited readers to drop a line with ideas. It linked to a dizzying animation showing how fast the planets spin by putting faces on each one; a mesmerizing homemade light made from a flask of mercury, a little neon, and some shaking; a digital collection on what the United States’ earliest restaurants teach about the country; and a science answer to the crucial Rick and Morty question of whether a lightsaber could really drill its way to Earth’s core.

Current stage: Anthropic has entered the public-filing phase, while the surrounding news cycle shows AI safety, energy supply, scientific publishing, and human-evolution research all moving toward harder tests.