Who’s really profiting from all the money pouring into AI?
This earnings season is revealing new details of big tech’s eye-popping spending on all things AI, and it shows no signs of slowing. But who is profiting from all this investment, and will it ever lead to profitable AI businesses?
Nvidia seems to be continuing to make a killing selling its AI computing hardware to all of the companies in the space. Unless there is a huge shift away from training ever larger AI models, its products are likely to be in demand.
Microsoft is making money selling OpenAI’s technology to customers via Azure, and they are planning long-term to meet demand. On Microsoft’s Q4 earnings call this week, executives said demand for AI computing from Azure boosted revenue, and signaled that large investments in data centers, and expensive GPUs will continue, laying out a 15 year timeline to build capacity, allowing them flexibility to respond to demand for AI services.
New reporting from The Information reveals that Microsoft is on track to make about $1 billion annually reselling OpenAI’s services (as part of their complicated partnership), but currently a quarter of that revenue is coming from one customer — TikTok, which could turn elsewhere for its AI computing.
Microsoft’s deal gives them access to OpenAI’s technology, and is rumored to include a hefty slice of their OpenAI profits until their investment is recouped.
Speaking of OpenAI, the company makes money selling Plus, Team, and Enterprise tiers of ChatGPT subscriptions, and by charging developers access to its API, which is estimated to generate several billion dollars per year. But OpenAI’s business depends upon expensive hardware, high energy costs and has to bankroll some of the highest paid roles in tech.
OpenAI has been busy spending Microsoft’s $10 billion investment on a quest to build artificial general intelligence, which may not be a thing that will ever actually exist. But industry observers are starting to question the fundamentals of OpenAI’s business and can’t figure out how it will continue to raise the cash it needs to power its research and development. Not to mention its ChatGPT service, which is incredibly expensive to operate.
OpenAI’s technology will be showing up on Apple iPhones this year as part of iOS18, but Apple isn’t paying them for the deal, raising more questions about how OpenAI will fund those increased costs.
Meta has been spending massively on AI research and hoarding expensive chips, with plans to spend between $35 billion and $40 billion on capital expenditures in 2024. But its AI spending hasn’t yielded much in the way of revenues yet, other than AI improvements to its advertising business.
Earnings reports from Meta later today and Amazon tomorrow may tell more of the story. Last quarter, AI was a key driver of big tech’s capex spending spree:
