Think about the dotcom bubble. We went from having all of these businesses that shot to the moon for valuation… and then it collapsed.
Amazon and Google and basically all of FAANG/MAMA beg to differ.
And basically every company still has a website and ecommerce in general is still massive.
The Dot Com bubble was brutal. But it was not a collapse of the industry. It was much more of a correction after so many companies blew up to sell services like “rate my dog” and many of the smaller ecommerce sites were absorbed by larger ones.
And that is likely what will happen with generative AI. I expect a MUCH bigger bloodbath for openai but anthropic seem to have scaled a lot better (although they also are going into their IPO with a massive data breach…). But expect most of the smaller companies to similarly get gutted.
But
Makes sense. It takes years to complete construction projects and budgets always overrun. AI is a fad for most companies trying to get into it.
Data Centers exploding… honestly have little to do with AI. No… not the way they explode in the UAE… metaphorical exploding. The reality is that data centers are GOOD money. Basically every company needs data hosting and offloading internal infrastructure is… honestly a really smart play. Same with the massive push for streaming of basically everything so that nobody owns anything and subscribe to everything. You want regional data centers and… this is how you get them.
Amazon and Google and basically all of FAANG/MAMA beg to differ.
But it was not a collapse of the industry.
I mean they don’t beg to differ at all and it absolutely collapsed. The result of which is the reason why FAANG as an acronym exists. FAANG is the resulting consolidation of the industry made possible by its widespread collapse.
It collapsed into… 4 or 5 of the most powerful companies in human history.
And that is the thing to understand. These bubbles are less about widespread collapse and more about consolidation. And it tends to be the smaller companies that suffer, not the biggies who were driving much of it to begin with.
That’s a bubble, quite literally. Collapsing into other companies. Do you think when people say bubble they mean that all companies related to LLMs will go under forever?
On the FAANG/MAAMA, he said explicitly some winners but a whole lot of losers, so you are agreeing with him. Of those titans, only Amazon and Google were arguably dot com darlings. Apple was pretty much left out of it and in bad shape, Microsoft did “ok” but was not really a darling of that bubble. Facebook, Netflix didn’t exist.
The data center explosion has everything to do with the AI boom. That’s the only change around the recent inflection point of growth. They were certainly prolific, but this is beyond. They are now demanding incredibly more power and cooling density than before. OpenAI by itself made purchasing commitments to the tune of 40 percent of the entire supply of ram production across all industries. Probably 75 to 80 percent of the recent plans would not have happened if not for the LLM craze.
Data centers existing makes sense, but this specific aggressive AI data center buildout (with special-purpose hardware) doesn’t: the two AI companies you mentioned, OpenAI and Anthropic, aren’t making a profit, and they don’t appear to have a viable path to one. OpenAI claims it’ll be wildly profitable in just a few years, but they don’t go into how.
“aren’t making a profit” gets into the mess that is book keeping and is a giant rabbit hole people actively avoid because it is just easier to get angry at stupidity rather than complex malfeasance.
But what makes something an “AI data center” outside of the branding?
The reality is that it is a shit ton of computers connected to a really fast internet connection. Preferably through a properly managed set of switches but you do you. And the reason that we still mostly use GPUs for “AI” rather than highly specialized hardware (although, nvidia DID just buy groq a few months back…) is for that reason. They might do linear algebra of quarter precision floats REALLY well but they also do linear algebra of single and double precision floats pretty well too. And the CPUs and mobos (that are mostly optimized for data movement to offload to said GPUs) are no slouches either.
Which is what most of these companies are planning for. openai is, arguably, really fucking stupid. Whereas anthropic have shown decent signs of “diversifying” as it were. And nvidia… if we lived in a world where they could get enough RAM I think they would be fine. As it stands… Jensen (and a LOT of people) are kinda fucked and I expect to see a hard pivot over the next 12 months.
Because if we banned ALL generative AI tomorrow? The people who think you can’t use a computer without installing litellm first are gonna be fucked. But everyone else will just put other workloads on there and be… “fine” is a strong word but they won’t go bankrupt. And the data centers themselves will still be incredibly valuable.
I wish GPUs in AI data centers (or worse, the ones purchased and not installed yet) were more general-purpose than they appear to be. That’s the part that makes them AI data centers: the optimized hardware.
I do agree things are complex. And I like reading about the intricacies of that complexity. The overall picture is still a pretty bad one, though.
Yes, there are some fairly revolutionary(-ish) chips. Those are few and far between because they tend to be hyper specialized. Inference but not training or only optimized for a very small input matrix (common for edge computing like cameras).
By and large? They really ARE “traditional” GPGPUs that are optimized to hell and back for vector operations and linear algebra. And a lot of the gains there come from multiplying their floating point performance by 2-4 (depending on if half or quarter precision). They aren’t as good for double precision as something optimized for it but basically only a very small subset of users need that. There will be no issues repurposing the hardware in these data centers.
And the rest is data movement which has always been the real problem.
I don’t think most companies will find much value in that though. I know that none of the infrastructure I work with uses heavy calculations, and if we tried to jam it in, we’d be making solutions looking for problems.
An email server doesn’t need a GPU, neither does a file server, or a website, or an e-commerce platform.
Suppose they could rent it out as supercomputers but I don’t think the return on cost is going to be that good.
Nice eval and analysis. I am finding it more and more obnoxious that lemmy doesn’t seem to like true debate. Take the down votes as a good sign that you struck a possible cord of reality.
FAANG exists because they have rock solid products with built in users. Their new business initiatives and decisions are woefully lacking. The new ideas aren’t bubbling up. Nvidia is the only one that isn’t staring into the abyss but their only products are iterative and once something revolutionary comes along they are sunk, like AMD v Intel these last 5 years.
Don’t think FAANG is a leader of anything except valuation. They are hopeless business leaders.
Amazon and Google and basically all of FAANG/MAMA beg to differ.
And basically every company still has a website and ecommerce in general is still massive.
The Dot Com bubble was brutal. But it was not a collapse of the industry. It was much more of a correction after so many companies blew up to sell services like “rate my dog” and many of the smaller ecommerce sites were absorbed by larger ones.
And that is likely what will happen with generative AI. I expect a MUCH bigger bloodbath for openai but anthropic seem to have scaled a lot better (although they also are going into their IPO with a massive data breach…). But expect most of the smaller companies to similarly get gutted.
But
Data Centers exploding… honestly have little to do with AI. No… not the way they explode in the UAE… metaphorical exploding. The reality is that data centers are GOOD money. Basically every company needs data hosting and offloading internal infrastructure is… honestly a really smart play. Same with the massive push for streaming of basically everything so that nobody owns anything and subscribe to everything. You want regional data centers and… this is how you get them.
Their comment literally mentions Google and Facebook as the survivors, so I’m not sure what good it does to include them in your argument.
I mean they don’t beg to differ at all and it absolutely collapsed. The result of which is the reason why FAANG as an acronym exists. FAANG is the resulting consolidation of the industry made possible by its widespread collapse.
It collapsed into… 4 or 5 of the most powerful companies in human history.
And that is the thing to understand. These bubbles are less about widespread collapse and more about consolidation. And it tends to be the smaller companies that suffer, not the biggies who were driving much of it to begin with.
That’s a bubble, quite literally. Collapsing into other companies. Do you think when people say bubble they mean that all companies related to LLMs will go under forever?
You’re just describing bubbles. That a few survived is not the strong point you think it is.
On the FAANG/MAAMA, he said explicitly some winners but a whole lot of losers, so you are agreeing with him. Of those titans, only Amazon and Google were arguably dot com darlings. Apple was pretty much left out of it and in bad shape, Microsoft did “ok” but was not really a darling of that bubble. Facebook, Netflix didn’t exist.
The data center explosion has everything to do with the AI boom. That’s the only change around the recent inflection point of growth. They were certainly prolific, but this is beyond. They are now demanding incredibly more power and cooling density than before. OpenAI by itself made purchasing commitments to the tune of 40 percent of the entire supply of ram production across all industries. Probably 75 to 80 percent of the recent plans would not have happened if not for the LLM craze.
Data centers existing makes sense, but this specific aggressive AI data center buildout (with special-purpose hardware) doesn’t: the two AI companies you mentioned, OpenAI and Anthropic, aren’t making a profit, and they don’t appear to have a viable path to one. OpenAI claims it’ll be wildly profitable in just a few years, but they don’t go into how.
“aren’t making a profit” gets into the mess that is book keeping and is a giant rabbit hole people actively avoid because it is just easier to get angry at stupidity rather than complex malfeasance.
But what makes something an “AI data center” outside of the branding?
The reality is that it is a shit ton of computers connected to a really fast internet connection. Preferably through a properly managed set of switches but you do you. And the reason that we still mostly use GPUs for “AI” rather than highly specialized hardware (although, nvidia DID just buy groq a few months back…) is for that reason. They might do linear algebra of quarter precision floats REALLY well but they also do linear algebra of single and double precision floats pretty well too. And the CPUs and mobos (that are mostly optimized for data movement to offload to said GPUs) are no slouches either.
Which is what most of these companies are planning for. openai is, arguably, really fucking stupid. Whereas anthropic have shown decent signs of “diversifying” as it were. And nvidia… if we lived in a world where they could get enough RAM I think they would be fine. As it stands… Jensen (and a LOT of people) are kinda fucked and I expect to see a hard pivot over the next 12 months.
Because if we banned ALL generative AI tomorrow? The people who think you can’t use a computer without installing litellm first are gonna be fucked. But everyone else will just put other workloads on there and be… “fine” is a strong word but they won’t go bankrupt. And the data centers themselves will still be incredibly valuable.
How is Anthropic diversifying?
I wish GPUs in AI data centers (or worse, the ones purchased and not installed yet) were more general-purpose than they appear to be. That’s the part that makes them AI data centers: the optimized hardware.
I do agree things are complex. And I like reading about the intricacies of that complexity. The overall picture is still a pretty bad one, though.
Ehhhh.
Yes, there are some fairly revolutionary(-ish) chips. Those are few and far between because they tend to be hyper specialized. Inference but not training or only optimized for a very small input matrix (common for edge computing like cameras).
By and large? They really ARE “traditional” GPGPUs that are optimized to hell and back for vector operations and linear algebra. And a lot of the gains there come from multiplying their floating point performance by 2-4 (depending on if half or quarter precision). They aren’t as good for double precision as something optimized for it but basically only a very small subset of users need that. There will be no issues repurposing the hardware in these data centers.
And the rest is data movement which has always been the real problem.
I don’t think most companies will find much value in that though. I know that none of the infrastructure I work with uses heavy calculations, and if we tried to jam it in, we’d be making solutions looking for problems.
An email server doesn’t need a GPU, neither does a file server, or a website, or an e-commerce platform.
Suppose they could rent it out as supercomputers but I don’t think the return on cost is going to be that good.
Nice eval and analysis. I am finding it more and more obnoxious that lemmy doesn’t seem to like true debate. Take the down votes as a good sign that you struck a possible cord of reality.
FAANG exists because they have rock solid products with built in users. Their new business initiatives and decisions are woefully lacking. The new ideas aren’t bubbling up. Nvidia is the only one that isn’t staring into the abyss but their only products are iterative and once something revolutionary comes along they are sunk, like AMD v Intel these last 5 years.
Don’t think FAANG is a leader of anything except valuation. They are hopeless business leaders.