71–80 of 249 posts
With the ASICs being so efficient at inference, I’m getting the feeling even the infrastructure investments are going to be severely outdated in a couple of years..
I was thinking any breakthrough in hardware (e.g. spintronics etc.), even if just partially effective, means all of this hardware would need to be replaced.
And, with this, the economic contagion begins. Before this step, we had non-existent money loaned to build datacenters powered by non-existent power infrastructure to support future load that will never exist, with guaranteed revenue from companies that do not, and will never be able to pay. Previously, this was propped up with circular revenue and investment. Now it's going to be propped up by dismantling the US tec…
That's the story for Wall Street. Oracle went on a huge run in the market, that it did not deserve, and they're going to attempt to hold on to as much of that gain as they can. Wall Street will applaud slashing jobs if you can give them a good reason for it (and sometimes with no reason at all).
Amazon, Meta, Apple, Nvidia, Microsoft, Google are producing $700+ billion per year in op income. They have nothing else to spend it on. They will continue to fund a gigantic AI build out.
It does end well eventually, just as the dotcom era build-out ended well over the following decades. It ends with continued US dominance. Yes, the dotcom era saw a crash. The build out continued.
Browsers? The US won.
Ecommerce? The US and China won.
Search? The US won. China didn't win the search war, Baidu was supposed to be a serious global challenger to Google, that was widely hailed across tech circles as a known fact - it was inevitable. Nope. Google won. And Europe never did field a competitor at all.
China didn't win the smartphone war. Apple and Google won globally. Apple has produced trillions of dollars in profit on the back of that fact.
Cloud? China didn't even come close globally. The US won without any challenge at all. Europe never showed up. But but but Hetzner.
China didn't win the GPU war. Nvidia has the market, at least for this decade. That will produce over one trillion dollars in profit in just the next five or six years.
China didn't win the AI war. It was OpenAI's ChatGPT that shocked the world and set everything in motion. Now the US has the top three models in GPT, Gemini and Claude. Europe? Not even in the running (although they'd like to believe they are).
This all ends with the US remaining on top, with China as its only real peer.
Europe? There is no Europe in the equation, just as there wasn't for cloud or mobile apps or smartphones or personal computing.
With the ASICs being so efficient at inference, I’m getting the feeling even the infrastructure investments are going to be severely outdated in a couple of years..
The economy is going to go into a deep recession for years with the current economic environment. There will be nobody left to spend money, just increasing corporate and ultra high wealth spend. The rest will be left with food and water. We are destroying currently valuable jobs to allocate resources to electricity and silicon, concentrating wealth into 10 companies pockets. I don't know what comes after, but when yo…
[flagged]
Smart people have sold their US backed holdings. I'm making bank on gold and the international market.
Earlier quoted context omitted.
ASICs only work for very small and heavily quantized models. Moreover, they are fixed function hardware, so whenever you have a new model, you have to throw the current chips away and design and buy new ones. That's like buying a new CPU every time a new OS version comes out.
Google’s training and running all their stuff on ASICs, seems to be working out well.
We will continue to be told this is because AI is driving efficiency instead of that it's driving CapEx and investors won't tolerate FCF collapsing.
I strongly believe AI will drive efficiency. I’m less certain about whether that’s already happening or will happen in the future and when in the future. However, I’ve already seen a CTO of a F100 company explicitly state that whether AI is driving efficiency or not, the capital investment, and more importantly, the promises of efficiency to investors will mean some people will be let go. Efficiency is output/input.…
With the ASICs being so efficient at inference, I’m getting the feeling even the infrastructure investments are going to be severely outdated in a couple of years..
ASICs only work for very small and heavily quantized models. Moreover, they are fixed function hardware, so whenever you have a new model, you have to throw the current chips away and design and buy new ones. That's like buying a new CPU every time a new OS version comes out.
Earlier quoted context omitted.
If it is 2030, compute and software will vastly reduce requirements. Nvidia has so much cash for R&D. You can bet they now have immense optimization and improvements in the pipeline, but why would they release anything groundbreaking right now? There are no real competitors on their heels. As Intel did for decades, they will likely dole out improvements, only when necessary to remain ahead. By 2030, I expect 10x impr…
When I read this comment all I see is: LLM at the edge - or close to it - will become available. And whoever provides the best eco-system across digital lifestyle and business wins. Oh... Apple? lol. Well that'd be funny wouldn't it. Oh and dont forget Apple got rid of its reliance on Intel too. No reason why this can't happen again.
And, with this, the economic contagion begins. Before this step, we had non-existent money loaned to build datacenters powered by non-existent power infrastructure to support future load that will never exist, with guaranteed revenue from companies that do not, and will never be able to pay. Previously, this was propped up with circular revenue and investment. Now it's going to be propped up by dismantling the US tec…
The economy is going to go into a deep recession for years with the current economic environment. There will be nobody left to spend money, just increasing corporate and ultra high wealth spend. The rest will be left with food and water. We are destroying currently valuable jobs to allocate resources to electricity and silicon, concentrating wealth into 10 companies pockets. I don't know what comes after, but when yo…
> I don't know what comes after, but when you combine this with the Iran war it's going to be closer to economic depression. I follow your extrapolation in the first part, but this is where you lost me. Why is the Iran war guaranteed to have long-term net negative consequences? It seems far too early to predict the outcome with any kind of certainty.
Combine that with the fact that the war is being led by a senile idiot who is unable to articulate a strategic purpose for the war in the first place and being prosecuted by someone who thinks that war crimes are aspirational, and you begin to understand that there is actually little prospect of this being resolved anytime soon.