The Company Building AI Just Raised Enough Money to Buy a Small Country
Money, power, and speed are reshaping AI this week — and not always in reassuring ways. OpenAI is reportedly worth more than Goldman Sachs. The electricity bill for running AI is becoming a political flashpoint. And a report out of Anthropic suggests AI is now helping build itself, twice as fast as it was a year ago. These three stories aren’t unrelated. They’re different angles on the same question: how fast is this moving, and who’s keeping up?
OpenAI Is Now Worth More Than Ford, Goldman Sachs, and Most Countries’ GDP
According to TechCrunch, OpenAI has raised an additional $11 billion in new funding, reportedly pushing its valuation to $1.25 trillion. A valuation is what investors collectively agree a company is worth — not necessarily what it earns, but what people believe it could be worth in the future. For context, Goldman Sachs, one of the most powerful banks on earth, is valued at roughly $200 billion.
That gap is striking. OpenAI doesn’t manufacture cars or manage trillions in assets. It builds AI software. The willingness of investors to pour this much money in reflects a belief that whoever controls the most capable AI systems will eventually control enormous amounts of economic value — the way Google once controlled search, or Amazon controls e-commerce infrastructure.
For everyday people, this matters in two ways. First, it signals that AI tools aren’t going anywhere — companies betting this much money on a technology tend to push hard to make it stick. Second, a company valued at $1.25 trillion faces enormous pressure to generate returns. That usually means higher prices, more aggressive product expansion, or both.
Why this matters: When a single AI company reaches this valuation, it has enough financial firepower to shape entire industries. That’s worth paying attention to, even if you’ve never used ChatGPT.
OpenAI raises $11 billion at $1.25 trillion valuation
AI Data Centers Are Pushing Power Grids to Their Limits
The Verge has been tracking an escalating tension between AI’s appetite for electricity and the communities living near its infrastructure. Data centers — the warehouse-sized buildings full of computers that run AI systems — consume staggering amounts of power. A single large facility can use as much electricity as a small city.
Think of it like this: running a conversation with an AI isn’t like loading a webpage. Every response requires thousands of calculations happening simultaneously, across racks of specialized hardware that run hot and need constant cooling. Scale that to millions of users, and the energy demand becomes genuinely industrial.
Residents near data center construction sites are raising alarms about rising electricity costs and strain on local grids. Politicians are starting to respond. This isn’t just an environmental story — it’s an infrastructure story. If your electricity bill goes up, or your neighborhood experiences more outages, there’s a real chance a data center nearby is part of the reason.
Why this matters: AI’s costs aren’t only financial. They show up on power bills and in community meetings, and those conversations are only getting louder.
Data centers consume so much electricity they’re creating problems for power grids
Anthropic’s AI Is Helping Build Better AI — and That’s Making Safety Researchers Nervous
According to The Verge, a new report on Anthropic — one of the leading AI safety companies — raises a genuinely unusual concern: Anthropic is reportedly using AI systems to help develop more powerful AI systems. The improvement rate has roughly doubled compared to a year ago.
This is sometimes called “recursive improvement” — when an AI helps make a better version of itself, which then helps make an even better version. The analogy is a factory that builds faster machines to build even faster machines. The concern isn’t that anything has gone wrong. The concern is that the speed of improvement makes it harder to spot problems before they’re baked into the next generation.
Safety researchers — people whose job is to identify risks in AI before it reaches the public — say the pace is outrunning their ability to thoroughly evaluate each new system. For regular people, this translates to a simple worry: if the people tasked with checking AI for problems can’t keep up with how fast it’s changing, who is?
Why this matters: Anthropic built its reputation on being the careful, safety-focused AI lab. If even they’re accelerating faster than their own researchers are comfortable with, that tells you something about where the whole field is headed.
AI-assisted models improving about twice as fast as they did a year ago
Also Happening in AI
On the technical side, Hugging Face released a maintenance update to its widely used Transformers library, patching security vulnerabilities — a quiet but important kind of upkeep for the tools developers use to build AI applications. Google DeepMind quietly updated its Gemini Flash model, offering faster responses and better handling of longer video clips, useful for anyone using Gemini for media-heavy tasks. Meanwhile, security researchers flagged hundreds of publicly exposed systems running Artex, an AI-powered hacking tool, according to TechCrunch — a reminder that AI security cuts both ways. And in an interesting wrinkle for the industry, Ramp’s spending data reportedly shows corporate AI budgets plateauing just as Anthropic approaches a potential public stock offering, suggesting that enthusiasm at the investor level doesn’t always match what companies are actually spending day to day.
What to Watch
The story underneath all of today’s news is speed — and who controls it. AI is improving faster, being funded at levels that were unthinkable three years ago, and consuming physical resources at a scale that’s starting to affect real infrastructure. Watch whether energy regulators start pushing back harder on data center expansion over the next few months. And pay close attention to whether Anthropic’s upcoming IPO proceeds smoothly — if the market is cooling on AI spending, that tension will show up in the offering price.