Comment by zmmmmm
15 hours ago
It feels like Anthropic is everybody's insurance policy against someone else winning the AI race. So you have Amazon, Google, Microsoft basically every major tech company pushing their own tech hard but simultaneously ensuring they have a survival level stake in Anthropic if they can't build or acquire their way to stay at frontier level performance themselves.
Maybe it was never really about maximizing the model technology as the ultimate end goal and far more about the business side and infrastructure.
The software will only improve for so long before it hits a wall. The best models were just a proxy for early mainstream market adoption, keeping your head above the water … plus some useful marketing hype about longshots for developing something bigger than LLMs (“AGI”).
People who work in tech are biased to obsess about the technical side and short term uptime/performance outrage. Despite that being mostly just standard immature market issues.
I find it interesting that Anthropic is in this position and not OpenAI. Where did OpenAI go wrong? Lack of focus and overambitious in some of their spending commitments?
Does not seem that complicated. OpenAI basically had to do a lock-in deal with Microsoft/Azure at the time, and they pioneered this circular funding hyperscaler deal structure so there were some rough edges.
Anthropic (all ex Open AI) knew the negatives of the deal, so they made a slightly better deal with AWS, not a full lock in. They also grounded it in hardware from the start, ie. being the flagship customer for Trainium, the flagship customer for external usage of TPU's.
Also, the fact that other tech leaders know very well how Sam Altman operates doesn’t help OpenAI secure deals with big tech.
Domain knowledge, expertise, is a big thing in tech, because code can be written fast if you know what to do, and so by having that expertise, building a frontier model is a matter of time and capex. Anthropic is founded by top ex-openAi, so they are not lacking in expertise, and are not attached to SamAlt. It’s an easy choice of who to finance.
Anthropic will win long term because big tech knows how much of a loud mouth sam is, how much of the piee he wants, he is more of a rival then some company they could use to grow. While Anthropic (even though they aren’t really good guys) seems more like a shared common good for the big tech then openAi, like linux corporate business deals version.
This seems like wild speculation that isn't even really true at all. OpenAI locked up all the compute capacity which is why Anthropic is struggling so badly with capacity to scale for demand. It's why Claude quality is plummeting and people are leaving in droves because the usage limits are pathetic and the API pricing structure is outrageous. All because they can't scale. So that's what this deal is about.
Isn’t the only thing OpenAI did was: throwing a half baked model out for the public to go ham on? I was at Google when they did this and we already had working LLMs internally, they just weren’t good enough to release without PR backlash. I don’t see why such a pithy “advantage” should have led to anything other than a moment in the spotlight? The “we have no moat and neither does OpenAI” essay was published very shortly afterwards.
If anything you ought to expect them to be behind, since they took the position of making all the mistakes first so others (who already had the same or better tech) didn’t have to.
> throwing a half baked model out for the public to go ham on?
I think that’s underselling their contribution, which I believe is mainly: it’s possible and this is what it looks like as a product. Until that time, nobody had figured out how to shape it as a product, and ChatGPT showed how to do that. Don’t forget that for a year or two they kept making headlines all the time with Dall.E and whatnot.
For me it seems like what happened after that is where the lack of focus started to hurt them: they realized that models themselves will be a commodity and have no moat, and that they needed to somehow build a network or something to keep pulling people back in. Sora was one such attempt, and it failed hard.
To me, enterprise / B2B seems like a much easier, obvious market to approach, but I don’t know a lot about B2C. But it seems like B2C was what OpenAI was going after.
[dead]
>Where did OpenAI go wrong?
OpenAI was Anthropic. Anyone involved in actually developing GPT jumped ship when Altman performed his coup.
[dead]