[MattGaiser]

Are they at risk as long as they can replicate the competitor quickly? Plenty of companies don't invest in R&D to avoid cannibalizing existing business. Google does the R&D, but just does not deploy it to customers immediately.

[sowbug]

In the case of a large company like Google, "immediately" can take years.

[kccqzy]

I think you overestimated it. Google's LaMDA is already ready and running, it's just that it's behind the corporate firewall available only to employees. Years? That's way too long for them to scale up the service and flip a switch to open it to the public.

[thefourthchime]

Large corporations are great at failing to exploit their own strengths because they are afraid of anything rocking how they make money.

I would be far more surprised if google reinvents itself using a transformer-based AI than I would some unknown company.

[sowbug]

Yeah, this is more where I'm coming from. Even if the product's technology were fully ready to launch, there are so many layers of stakeholders, existing overlapping product lines, branding questions, and, yes, pure politics to overcome. That's all in addition to the classic Christensen innovator's dilemma whether the company even wants to disrupt its own business model.

It's a lot more than removing an ACL on a server.

[visarga]

Well, chatGPT demo managed to put Azure GPUs on their knees, and that was just a million users. When you got a billion users you need a smaller model or many more datacentres.

[nailer]

> quickly

Google’s major products were either built or acquired in the mid 2000s.