Anthropic's IPO, COBOL's Crisis, and the AGI Bet
October 26, 2026Three AI companies dominated financial news this week, and somehow the most interesting story is about sixty-year-old computer code.
Anthropic Goes Public — and Gets Unusually Honest About It
Anthropic filed its S-1 (the disclosure document every company must submit before selling shares to the public) with the SEC on Monday, preparing to list on the Nasdaq under the ticker "ANT." The filing reveals a company with soaring revenue but deepening losses, which is normal for this stage, and an unusual governance structure that keeps founders in control even after the IPO — a deliberate choice to prevent short-term shareholder pressure from overriding its AI safety mission. What's less typical is the candor: the risk factors section reads almost like a company warning you not to invest, acknowledging that Anthropic's own AI could cause serious harm and that the business model depends on assumptions about the future of intelligence that could simply be wrong. For non-technical professionals, the practical takeaway is that Claude's maker is about to become a public company, which means quarterly earnings calls, analyst scrutiny, and pressure to show you can turn a massive research operation into a profitable business. That tension between "we're trying to save humanity" and "we need to hit our numbers" is real — and the first time Claude makes a decision that kills a deal, there'll be an earnings call question about it.
https://techcrunch.com/2026/10/26/anthropic-officially-files-for-its-public-debut/
Anthropic Is Beating OpenAI to Wall Street
The Verge's take on the same filing is worth reading separately because it zeroes in on something competitive: Anthropic is moving faster toward an IPO than OpenAI, its better-known rival. OpenAI has been restructuring its corporate governance for over a year, and that process has repeatedly delayed any public offering. Anthropic, quieter and less culturally prominent, just threaded that needle first. If you work in enterprise software sales, procurement, or consulting and you're still treating Anthropic as the scrappy alternative to OpenAI, stop calling them that. They're not.
https://www.theverge.com/ai-artificial-intelligence/1018476/anthropic-ipo-details-s1-filed-sec
Tesla Decides Cars Are Just the Hardware
Tesla, having lost its dominance in electric vehicle sales to Chinese manufacturers and legacy automakers, is now betting that the real business was always the software layer on top. The plan is to sell autonomy as a subscription — meaning you pay monthly for your car to drive itself, the way you pay for Netflix — and that recurring revenue would deliver profit margins that physical car sales never could. If you're in fleet management, your cost model just became someone else's pricing decision. Tesla has been promising full self-driving for a decade and keeps moving the goalposts, so the credibility question is real. But the underlying model — hardware as a distribution channel for software revenue — is how the whole industry is moving. Apple's been running this playbook since the App Store. John Deere has been trying it with tractors for years.
https://arstechnica.com/cars/2026/10/tesla-wants-ai-and-software-to-be-its-next-profit-engine/
The Quiet Crisis: Ancient Code Running Everything You Depend On
COBOL is a programming language from 1959. It runs the IRS, most major banks, airline reservation systems, and significant chunks of government infrastructure worldwide. It's not a legacy curiosity — it processes an estimated $3 trillion in daily commerce. The problem is that the people who know how to maintain it are retiring, and almost nobody young is learning it. OpenAI has been positioning its models as a potential translation layer, able to read COBOL and rewrite it in modern languages that thousands of current developers understand. The risk is enormous: these systems are so old and so intertwined that even small translation errors could cause the kind of failures that make headlines and cost billions. This is not a theoretical future problem. It is happening now, and the organizations sitting on these systems are starting to make expensive bets on whether AI can safely modernize them.
Kevin Roose Is Convinced AGI Is Close. He's Not Alone.
Kevin Roose, a journalist who has covered AI longer than most, has written a book arguing that artificial general intelligence (AGI — meaning AI that can do essentially any intellectual task a human can, rather than the narrow tools we have today) is arriving faster than most people outside the industry believe. He joined The Verge's podcast to make the case, and the interview is worth an hour of your time if you want to understand why the mood inside AI labs has shifted from "eventually" to "soon." Roose might be wrong on the timeline and still be right that people are making irreversible bets as if he's correct — which is almost worse. The CEOs of Microsoft, Google, and Amazon are all publicly saying AGI is close. You can think they're wrong, but they're allocating capital like they're right.
The most consequential AI story of the next five years might not be a new model release — it might be what happens when a publicly traded Anthropic has to explain a safety incident to shareholders on a quarterly earnings call.
Stop staring at the blank ChatGPT box
The Plain-English AI Prompt Pack: 100 copy-paste prompts for real work — emails, meetings, marketing, managing people. Works with any chatbot.
Get the Prompt Pack →Get the daily brief
Plain-English AI news for people with real jobs. Free, three minutes a day.