| Today's Takeaway is by Julie Hyman, Anchor.
There's a spending frenzy around AI, and this earnings season we've been getting more details on the pace of that outlay.
Meta raised its capital expenditures forecast to the range of $37 billion to $40 billion. Microsoft spent $19 billion last quarter, including server farm leases. Amazon spent $30 billion so far this year and says it'll kick it up a notch to close out 2024. Apple may be zagging, but clearly has a fresh AI focus of its own.
We've been told that generative AI is the next smartphone, the next internet — a transformational technology. That's why these companies have ramped up spending in a generative AI-building arms race.
They're spending because they believe in it, of course.
But is the spending also influencing that belief? Are the stakes now so high that even if the promise of generative AI doesn't meet those lofty expectations, these companies, and their clients, will try to shoehorn it into products and processes where it's not really needed?
History is littered with "transformational" technologies that didn't quite make the grade.
Digital assistants, for example, like Amazon's Alexa. The Wall Street Journal recently quoted a former senior employee as saying, "We worried we've hired 10,000 people and we've built a smart timer." Apple was reportedly spending $1 billion annually to develop a supposedly industry-changing vehicle before abandoning that effort earlier this year.
Of course, a moonshot effort like the car was a sideline to Apple's main business and hardly a lottery ticket for the company. It seems more so for Alphabet, Meta, Microsoft, even Tesla, which are putting AI front and center in their pitch to investors. They're motivated to make it work. And make it work quickly. Sequoia Capital partner David Cahn called the infrastructure build a "cycle of competitive escalation" in a recent piece, positing that the hyperscalers would move much more slowly, and perhaps rationally, if not on each other's heels. The ensuing digestion period has already likely begun, if investors' impatient reaction to Microsoft's earnings is any indication. Meta, on the other hand, "continues to earn the right to spend big on GenAI," wrote JPMorgan's Doug Anmuth in reaction to its numbers. As he pointed out, "Management continues to expect Meta AI to become the most used AI assistant by the end of 2024, & open source Llama 3.1 should support a faster pace of innovation, better social/advertiser experiences, & cost benefits." Is that transformational? Expect to continue to hear that it is, or will be. It has to — because they're writing the checks. | 
