Hacker News Reader

Comment by curt15 | Hacker News Reader

Is Google also deep in the hole with Gemini, or is it just OpenAI that can't make ends meet?

Also important to remember that Google is years ahead of most other AI shops in that they're running on custom silicon. This makes their inference (and maybe training) cheaper then almost any other company. People don't realize this when compared to OpenAI/Anthropic where most folks are utilizing NVIDIA GPUs, Google is completely different in that aspect with their custom TPU platform.

> Also important to remember that Google is years ahead of most other AI shops in that they're running on custom silicon.

Not just the chips, Google's entire datacenter setup seems much more mature (e.g. liquid cooling, networking, etc.). I saw some video of new Amazon datacenter (https://www.youtube.com/watch?v=vnGC4YS36gU) and it looks like a bunch of server racks in a warehouse.

Google’s datacenters are excellent, from what I’ve seen in my career. They genuinely had so many amazingly talented SMEs pushing boundaries for decades without executive intervention or deterrence, and that’s paid dividends in the subsequent tenure under Pichai and external shareholders (in that they have “infinite” runway and cash reserves to squander on moonshots before risking the company’s core businesses). That said, nothing lasts forever, and if their foray into LLMs don’t pay off, their shareholders are going to be pissed.

And not just pushing the boundaries, working with the HW vendors to define them, asking for features and design elements that others don't really even see the point of.

Anthropic uses TPUs as well as nvidia. Compiler bugs in the tooling around the platform caused most of their quality issues and customer churn this year, but I think they've since announced a big expansion in use:

https://www.anthropic.com/engineering/a-postmortem-of-three-...

Depends what you mean by deep in the hole. Given how liberal Gemini is with free access it hardly seems like moneymaking is the end goal. Amongst AI labs, I've personally found Deepmind to be consistently the least sensationalist and least concerned with marketing. I think they're doing a great job of commoditising their complement.

https://gwern.net/complement

Google has enough income that they can self fund Gemini at a loss on whatever timespan they want. OpenAI does not.

ChatGPT has replaced Google Search for many people. If ChatGPT can also replace Gmail, Maps, and YouTube (by generating content for each of these), Google is done.

Replace Gmail? Meaning instead of real emails from my actual contacts, I’ll read AI generated emails?

Why not? I already use LLMs to generate summaries of my emails, search emails, and perform a host of email related operations as do many people. No reason OpenAI couldn't release their own email service that tightly integrates their LLM with email directly.

What's interesting about your comment is that back in 2004 people were like "Google releasing an email service? Why on Earth would they do that, that makes no sense!"

Yes but ChatGPT has not replaced Google Search at a profit. ChatGPT loses revenue per user. Certainly the hope is that at some point they become profitable but the concern is that might not happen.

Google has the revenue to cover their spending and then some.

It cannot be stressed or even succinctly elaborated how far ahead Google is compared to every other organization doing anything remotely like this, anywhere on the planet. Practically speaking there is no "catching up" to Google; you can only hope to capitalize on any mistakes that they make.

Google has a massive ad revenue that’s funding its AI habit. The difference between Google and OpenAI is that of a rich guy buying an expensive car that he can easily afford and a working class guy having a mid life crisis.

Reply on news.ycombinator.com