Google outspent OpenAI with $85 billion in capital expenditure in 2025, but with one crucial difference: that spending was funded by existing profitable businesses. Search advertising alone generates over $200 billion annually. YouTube prints money. Google Workspace has 3 billion users. Its AI investments are layered on top of a cash-generating mountain, not instead of one. Google already has the income that OpenAI is shakily projecting it will one day earn.
Google has successfully integrated AI into Search without cannibalising core advertising revenue. Query volumes are up. Monetization is stable. The existential threat everyone forecast hasn’t materialised. In a reversal of the predicted death blow, Google is using AI to strengthen its position.
Infrastructure matters. Vertical integration matters. Existing cash flow matters. A clear path from innovation to profit doesn’t have to be immediate or completely free from risk, but it needs to be probable. And that’s not a word that springs to mind when you examine Sam Altman’s position.
OpenAI will probably generate an astounding $30 billion this year, but that number needs to hit $200 billion four years from now to ensure profitability and survival. And none of the usual four growth vectors suggest this is possible, let alone likely.
Market expansion is the obvious play. But OpenAI is already global and well-penetrated. And expansion in AI doesn’t work like traditional market development because each new user in each new market adds computing costs. Rather than reduce the burn, user growth compounds it.
Clearly, Altman needs to increase prices to make those users more profitable. But he can’t raise them much beyond the $20 a month already charged. He only converts 5% of users into paying subscribers at the level. If anything, Altman is heading in the opposite pricing direction with ChatGPT Go launching at $8 a month.
He could diversify, which is happening. OpenAI has launched Sora for video. A browser called Atlas. He’s working on consumer hardware with Jony Ive’s design firm.
But it’s no fix. These new products require more R&D, more computing, more capital. OpenAI likely spent $13 billion on R&D last year. Product development isn’t the path to profitability: it’s another spade making that huge hole huger.
So, we arrive at last week’s desperate pivot into carrying advertising. OpenAI has announced it will test ads at the bottom of responses across ChatGPT’s free and Go tiers. It’s very much a last resort. Not my words. Sam Altman’s, back in 2024.
Everybody hates advertising until they lose their cat, as David Droga once sagely observed. Sam Altman’s kitty is currently wandering the streets, and he’s out there with a flashlight and a bag of treats.
