The investment in AI is ~90% R&D. Maybe more. It's fine to argue that the research will not pan out, but this article is entirely criticism of an R&D investment pattern.
The current zeitgeist reminds me a little of when Musk acquired twitter and people kept saying he would go bankrupt - he paid way too much for it - $44B. An idiot and also the likely world's first trillionaire.
> SpaceX and Anthropic’s Compute Deal Shows That There’s Little Demand Outside of Anthropic and OpenAI For GPUs
I almost stopped reading here. I get that he's captured a paying audience that had their brains largely turned off as long as the AI hate keeps spewing, but he's not this dumb. The world is drowning in demand for GPUs and memory for all purposes. There are dozens of reasons that Musk would lease this capacity to Anthropic; it's not like they can just spin up a hyperscale public cloud service offering overnight with this capacity. Musk shot himself in the foot trying to compete with better players in the agent space and needs a way to justify the hardware, this is like the only way to do it.
> Where is the demand coming from if it’s not OpenAI and Anthropic?
It is OpenAI and Anthropic, and their competitors, and they can't use it yet because it's not built yet, and they are serving customer demand right now.
That's investing, which is not the same pitch Ed is making repeatedly, that there is absolutely no value to this tech and everyone is lying about it. Saying it won't pay off to the level of investment is one thing, and may be true, but that is not his line.
Thats not investing - firms are paying out to llm producers in the form of operating expenses. What asset are they acquiring that deems it to be a investment? lmao
Laying people off offsets some of those expenses. Eventually revenue has to climb.
The reality is many of these tech firms have reached the peak of what they can offer - Meta and Google still have some ways to go with existing assets pre-LLMs that can generate lots more cash flows. So the sobering truth eventually will be that this tool seems really great but its not going to make the firm seem more productive in terms of its financials, beyond efficiency-generated growth which is finite.
And whether you like it or not, thats the only thing that matters. The firm exists to maximize wealth of its stockholders.
Most firms would consider experimenting with efficiency gains "R&D", which is investment. The AI companies under scrutiny here are gathering up as much compute as they can because people and companies are paying for it, and anybody pretending they're doing that for no reason is covering their eyes and ears.
> Laying people off offsets some of those expenses.
Layoffs are independent of AI. The impact of AI is that firms can do more with less, they are doing more with less, and when the economy supports them having more, they'll do more, not less.
> The firm exists to maximize wealth of its stockholders.
Deeply tired of hearing this type of argument. Your grandma is a stockholder. Your 401k is a stockholder. The Stockholders aren't some alien leech race. We've structured the economy so that the average person doesn't have to grow their own food; the result is stockholders, and it's not a scary word.
Ed is posturing to his explicitly anti-AI audience. Read any of his recent work on this topic and it's normal to feel embarrassed for agreeing with any of it.
The investment in AI is ~90% R&D. Maybe more. It's fine to argue that the research will not pan out, but this article is entirely criticism of an R&D investment pattern.
The current zeitgeist reminds me a little of when Musk acquired twitter and people kept saying he would go bankrupt - he paid way too much for it - $44B. An idiot and also the likely world's first trillionaire.
> SpaceX and Anthropic’s Compute Deal Shows That There’s Little Demand Outside of Anthropic and OpenAI For GPUs
I almost stopped reading here. I get that he's captured a paying audience that had their brains largely turned off as long as the AI hate keeps spewing, but he's not this dumb. The world is drowning in demand for GPUs and memory for all purposes. There are dozens of reasons that Musk would lease this capacity to Anthropic; it's not like they can just spin up a hyperscale public cloud service offering overnight with this capacity. Musk shot himself in the foot trying to compete with better players in the agent space and needs a way to justify the hardware, this is like the only way to do it.
> Where is the demand coming from if it’s not OpenAI and Anthropic?
It is OpenAI and Anthropic, and their competitors, and they can't use it yet because it's not built yet, and they are serving customer demand right now.
The world is drowning with firms who are experimenting in search of being able to create value by paying for tokens.
That's not quite the same thing as paying for a product with absolute certainty each purchase comes with it a benefit.
That's investing, which is not the same pitch Ed is making repeatedly, that there is absolutely no value to this tech and everyone is lying about it. Saying it won't pay off to the level of investment is one thing, and may be true, but that is not his line.
Thats not investing - firms are paying out to llm producers in the form of operating expenses. What asset are they acquiring that deems it to be a investment? lmao
Laying people off offsets some of those expenses. Eventually revenue has to climb.
The reality is many of these tech firms have reached the peak of what they can offer - Meta and Google still have some ways to go with existing assets pre-LLMs that can generate lots more cash flows. So the sobering truth eventually will be that this tool seems really great but its not going to make the firm seem more productive in terms of its financials, beyond efficiency-generated growth which is finite.
And whether you like it or not, thats the only thing that matters. The firm exists to maximize wealth of its stockholders.
Most firms would consider experimenting with efficiency gains "R&D", which is investment. The AI companies under scrutiny here are gathering up as much compute as they can because people and companies are paying for it, and anybody pretending they're doing that for no reason is covering their eyes and ears.
> Laying people off offsets some of those expenses.
Layoffs are independent of AI. The impact of AI is that firms can do more with less, they are doing more with less, and when the economy supports them having more, they'll do more, not less.
> The firm exists to maximize wealth of its stockholders.
Deeply tired of hearing this type of argument. Your grandma is a stockholder. Your 401k is a stockholder. The Stockholders aren't some alien leech race. We've structured the economy so that the average person doesn't have to grow their own food; the result is stockholders, and it's not a scary word.
Ed is posturing to his explicitly anti-AI audience. Read any of his recent work on this topic and it's normal to feel embarrassed for agreeing with any of it.
> in the foot trying to compete with better players in the agent space
We honestly don't know who the better players are. Stuff is changing quite a bit month by month.
It might be hard to rank the players exactly but nobody is writing significant amounts of code with a Grok subscription.