Hacker News Reader

Hacker News

Comment by nothrowaways | Hacker News Reader

Full thread

nothrowaways

> Meta is falling behind in the AI race

How so?

View on HN

Same way companies fell behind during the original "dot com" era: By not burning money quickly enough.

Disclosure: I work for one of the companies that fell behind.

Was that the real reason companues failed? By not burning money, instead of burning through it too fast?

Sorry for the late reply. I was being somewhat sarcastic, or maybe not. During the "dot com" era, investors were throwing money at anybody willing to take it, and a start-up was questioned about whether it was maintaining a sufficient "burn rate." In hindsight, it was a speculative bubble.

The company that I work for survived that era by not doing those things.

I guess, but if you invested $1 million into 100 companies in 2000 and one of them was Amazon (and other 99 went to zero) your investments combined would have had an annual growth of 24% over the past 25 years.

Someone in 2000 would not have done your proposed strategy unless they could name 100 .com like tech companies from 1975 that, if you had invested $1 million dollars in, would have returned 24% over the previous 25 years.

If the idea is just look at what worked 25 years ago, the 2000 investor would probably have followed a different approach.

Their latest Llama is a couple steps behind the frontier. Google, Anthropic, and OpenAI are all well ahead.

They missed the boat on LLMs and have been playing catch up.

Why should they be in that market at all? I don't understand why we've got to this place where every big tech company needs to have it's fingers in every pie or "they're falling behind". AI / LLMs look to be on their way to being infrastructure more than product in and of themselves. A patient huge company could build on top of what others are doing and wait for the market to settle, then buy for vertical integration.

They're not even in the same ballpark as the SOTA models, and they're not even in the same ballpark as the SOTA open source models, which was supposed to be their niche.

But even if they were, it's not immediately clear how they plan to make any money with having an open source model. So far, their applications of their AI, i.e. fake AI characters within their social media, are some of the dumbest ideas I've ever heard of.

Deepseek is a solid competitor in their niche, the open-source LLM niche. Llama 4 didn't impress, and so they are very vulnerable even in the free niche they carved out for themselves

How much revenue do they make with it?

I’m not sure revenue is the best measurement of success.

How would you even begin to calculate that? They use AI to make their products more valuable to advertisers, but there's no way to say how much their revenue would decline if they suddenly eliminated AI.

Reply on news.ycombinator.com