DooDooLamb News
The AI industry’s race for profits is now existential
Brief published April 11, 2026 · Original source published April 9, 2026
Original reporting by Nilay Patel at theverge.com.
Automated brief. Verify important details at the original source.
AI Companies Discover Money Doesn't Grow on GPU Trees
The artificial intelligence industry has finally stumbled upon a shocking revelation that would make even the most junior MBA student proud: at some point, you actually have to make money. This earth-shattering discovery comes as OpenAI and Anthropic frantically flip through business textbooks, looking for the chapter titled "How to Turn Billions in Funding Into Actual Revenue."
According to a new report from The Verge, the AI darlings of Silicon Valley are approaching what economists call "the monetization cliff" (a technical term meaning "oh shit, we spent all the money"). This development has sent shockwaves through an industry that has spent the last three years operating under the revolutionary business model of "step 1: build ChatGPT, step 2: ???, step 3: profit."
The timing couldn't be more perfect. Just as OpenAI burns through approximately $700,000 per day keeping ChatGPT running, they've discovered that most people aren't actually willing to pay $20 per month for the privilege of asking an AI whether a hot dog is a sandwich. Meanwhile, Anthropic continues to insist that Claude is different because it's "constitutional" (not in the legal sense, but in the "we wrote down some rules and hope the robot follows them" sense). Both companies are now learning that venture capitalists, despite their reputation for infinite patience, occasionally expect something called "returns on investment."
The existential crisis deepens when you consider what these companies actually sell. OpenAI's primary revenue stream appears to be ChatGPT Plus subscriptions and API access, which is essentially charging people to talk to a very expensive autocomplete feature that sometimes hallucinates entire Wikipedia articles. Their API pricing follows the time-honored tradition of cloud computing: make it cheap enough that developers integrate it everywhere, then slowly turn up the heat like a frog in a pot until enterprise customers are paying thousands per month for what amounts to fancy text generation.
Anthropic, meanwhile, has positioned itself as the "safety-first" AI company, which in practice means they charge similar prices while spending considerably more time explaining why their AI won't help you write a strongly-worded email to your landlord. Their business model appears to be "what if OpenAI, but with more PhD dissertations about alignment?" This strategy has proven remarkably effective at raising funding from investors who want to feel good about potentially destroying humanity.
The broader AI ecosystem faces similar challenges. Companies like Stability AI burned through funding faster than a teenager with a credit card, while Midjourney discovered that people will pay for AI-generated art right up until they realize they can't actually own the copyright to images of "cyberpunk cats in the style of Van Gogh." The entire industry operates on the assumption that artificial general intelligence (AGI) is just around the corner, and once it arrives, money will flow like water from a broken dam. This is roughly equivalent to planning your retirement around winning the lottery.
What makes this monetization scramble particularly entertaining is watching companies try to justify their eye-watering valuations. OpenAI's recent funding round valued the company at $157 billion, which means investors believe a chatbot company is worth more than most countries' GDP. To put this in perspective, that valuation suggests every human on Earth should pay OpenAI approximately $20 just for existing. The math checks out if you assume every person will eventually need an AI assistant to help them write passive-aggressive Slack messages.
The infrastructure costs alone should have been a red flag. Running large language models requires enough electricity to power small nations, and the specialized chips needed cost more than luxury cars. It's like building a business around giving away free rides in Formula 1 cars and hoping to make it up in volume. NVIDIA, the company that makes these chips, has become the accidental winner in this scenario, watching AI companies line up to buy their H100 GPUs like kids queuing for the latest iPhone.
Here's the uncomfortable truth that nobody in Silicon Valley wants to acknowledge: most AI applications are solutions looking for problems. The technology is genuinely impressive, but impressive doesn't automatically translate to profitable. We've reached the point where companies are adding AI features to everything from toothbrushes to tax software, not because consumers demanded it, but because "AI-powered" is the only way to get investor attention. The result is a landscape filled with products that use million-dollar models to solve ten-dollar problems.
The path to profitability isn't impossible, but it requires something the AI industry has steadfastly avoided: boring, sustainable business models. Instead of chasing AGI moonshots, these companies might need to focus on specific, measurable problems that customers actually pay to solve. Revolutionary technology, it turns out, still needs to follow the ancient business principle of charging more for your product than it costs to make.
As 2024 unfolds, we're about to discover which AI companies can successfully navigate the transition from "move fast and break things" to "move fast and pay rent." Place your bets accordingly, and remember: in the end, even artificial intelligence can't escape the very human problem of making the numbers add up.