Gurley calls foundational model companies' hundred-million-dollar annual burn rates poor capital allocation but possibly unavoidable competitive traps.
“200,000,000 a year? There's no way that's high quality capital allocation from my point of view. There's no chance. But maybe they don't have the alternative.”
Gurley warns AI companies trading equity for server capacity don't recognize true COGS and get warped decision-making.
“They're trading equity for server capacity. And so those companies are running on a day to day basis with a high COGS, but they don't know it because they're not paying cash for it.”
Gurley suspects some AI companies are reselling tokens below cost, creating unsustainable growth.
“I suspect there are companies that are selling, reselling tokens from Amazon or Anthropic or whoever at a price lower than they're paying for them. And which looks like growth, but it is unsustainable.”