Yeah, that’ll happen.
While AI bubble talk fills the air these days, with fears of overinvestment that could pop at any time, something of a contradiction is brewing on the ground: Companies like Google and OpenAI can barely build infrastructure fast enough to fill their AI needs.
During an all-hands meeting earlier this month, Google’s AI infrastructure head Amin Vahdat told employees that the company must double its serving capacity every six months to meet demand for artificial intelligence services, reports CNBC. The comments show a rare look at what Google executives are telling its own employees internally. Vahdat, a vice president at Google Cloud, presented slides to its employees showing the company needs to scale “the next 1000x in 4-5 years.”
While a thousandfold increase in compute capacity sounds ambitious by itself, Vahdat noted some key constraints: Google needs to be able to deliver this increase in capability, compute, and storage networking “for essentially the same cost and increasingly, the same power, the same energy level,” he told employees during the meeting. “It won’t be easy but through collaboration and co-design, we’re going to get there.”
Doubling every year would be crazy unsustainable at Google scale, but to tell employees you need to do so every 6 months seems like a fever dream of someone that doesn’t understand what they are asking (or knows they aren’t responsible for actually doing it). The old “just throw it over the fence” approach that corporate kool-aid drinkers love because they can take credit and shift blame.
so… it’ll collapse in 6 months?
Well that’s totally sustainable.
If someone says 10x it’s probably bullshit. 100x it’s definitely bullshit. I don’t even have a term for 1000x
Certified bullshit
Schweinerei.
collapse already
Gaslighting. Processor capacity doubles every 2 years according to Moore’s law which at best can be energy neutral. All other increases can only come from proportional energy increases. So instead of needing 1000x the energy, at best it is 250x.
the next 1000x in 4-5 years
At the risk of stating the obvious, Ars is working backwards from this metric to get their headline “double every 6 months.” 2^10 = 1024, to get that number in 5 years means doubling every half-year.
But Google didn’t set incremental 6-month deadlines for 5 years straight, they set a single 5-year deadline. Because in 6 months shareholders can call their bluff quite easily, but in 5 years they’re hoping everyone is A) distracted by some new disaster, or B) there’s a new tech hype cycle they can push. They’re trying to stall the bubble popping by pointing to a nebulous future where they magically scale to infinity, and hoping we all forget that they ever made this made this claim.
