14X. That’s how much DeepSeek’s revenue spiked last quarter, making its CEO one of the world’s richest people overnight. Meanwhile, Nvidia’s Jensen Huang casually announced we’ve “achieved AGI,” and the AI industry collectively shrugged because no one can agree on what that even means.
This is the state of artificial intelligence in 2025: CEOs throwing around terms like confetti at a parade, investors losing their minds over revenue multiples, and the rest of us trying to figure out if we’re living through a technological revolution or the world’s most expensive marketing campaign.
The AGI Definition Problem
Huang’s declaration would be monumental if anyone could pin down what AGI actually is. Ask ten AI researchers and you’ll get eleven definitions. Is it human-level intelligence? Superhuman performance across all tasks? The ability to learn anything a human can learn? A system that can make you a decent cup of coffee without burning down your kitchen?
The ambiguity isn’t accidental. It’s strategically useful. When the goalposts are mounted on wheels, you can always claim you’ve scored.
Look at what’s actually happening in the market. Character.AI just banned teens from their platform amid lawsuits and regulatory pressure. That’s not the move of a company confident in their technology’s maturity. That’s damage control from a product that can’t reliably avoid harmful outputs when talking to vulnerable users.
Follow the Money, Not the Hype
The real story isn’t in the AGI declarations. It’s in the numbers. Alexandr Wang, a college dropout, just closed a $14.3 billion deal with Meta for AI infrastructure. Siemens is betting on Germany’s industrial data sets for their AI push. CEOs are using “one number” to decide how many employees they still need.
These aren’t the actions of people who think AGI has arrived. These are the moves of people positioning themselves for what comes next, whatever that is.
DeepSeek’s 14X revenue spike tells you everything about where we actually are: early innings of commercialization, not the endgame of artificial intelligence. Companies are making money hand over fist not because they’ve solved intelligence, but because they’ve solved specific, valuable problems that businesses will pay for.
What We Actually Have
Strip away the terminology warfare and here’s what exists: AI systems that are genuinely useful for narrow tasks, occasionally impressive at broader ones, and consistently unreliable when you need them to be dependable.
Can current AI write code? Yes, often well. Can it reason through complex problems? Sometimes. Can it do so consistently, safely, and without hallucinating facts? Not remotely close.
The gap between “achieved AGI” and “banned teens from using our chatbot because we can’t control what it says” is the size of the Grand Canyon. Both of these things happened in the same news cycle.
The Real Question
Maybe the AGI debate is the wrong conversation entirely. Maybe we should be asking: do we need AGI for AI to be transformative?
The evidence suggests no. Companies are already restructuring around AI capabilities that fall well short of general intelligence. The “one number” Fortune mentioned that CEOs are using for workforce decisions? That’s not based on AGI. That’s based on narrow AI tools that can automate specific workflows.
The industrial data sets Siemens is excited about? Those feed specialized models for manufacturing optimization, not general-purpose thinking machines.
Wang’s massive Meta deal? Infrastructure for training and deploying models that are powerful but decidedly not general intelligence.
Where This Leaves Us
Huang’s AGI claim is either premature, definitionally creative, or marketing genius depending on your perspective. Probably all three.
What’s certain is that the AI industry has a credibility problem. When the same technology is simultaneously declared as achieved general intelligence and too dangerous to let teenagers use unsupervised, something doesn’t add up.
The money flowing into AI is real. The capabilities are real. The business transformations are real. But the gap between what’s being claimed and what’s being delivered grows wider every time a CEO makes a bold declaration that the rest of the industry immediately disputes.
We haven’t achieved AGI. We’ve achieved something else: AI systems powerful enough to be genuinely useful and genuinely concerning, but not powerful enough to deserve the god-like terminology we keep slapping on them. That’s the actual story, even if it’s less exciting than Huang’s headline.
The sooner we get honest about where we actually are, the sooner we can have productive conversations about where we’re going and what guardrails we need along the way.
đź•’ Published: