There are a lot of videos on this renaissance. Maybe an advance in electronics will make it worthwhile in hardware.

One area ternary is investigated is LLMs/classification. Bitnet was the pioneer model…

no no no… just for today… this is shit on America day.

if just one obscure one, that doesn’t deserve to be this obscure: KCLR-future funk chant. https://www.youtube.com/watch?v=1Sk75tWiEo8

a 2080S from 2019, a high end SKU, would struggle with many modern games at 1440p and higher. A profession streamer would be unlikely to use a 2080S.

On one hand a 2080s would still be good at doing what it was doing 6 years ago. If there are new needs, and unlimited power availability, then a new card in addition to whatever AI workload the 6 year old GPU can do in addition to the new card makes sense… if that card still works. Selling your 2080s or whatever old card, does mean a fairly steep loss compared to original price, but 6 year depreciation schedule is ok… IF the cards are still working 6 years later.

$3m NVL72 systems are a bit different, as one out of 72 cards burning out can screw up whole system, and datacenter power structure and expertise requirements, would have low resale value, though I assume the cards can be ripped out and sold individually.

They very much can be expected to engage in what is essentially accounting fraud.

Oracle this week “proudly boasted” that they get 30% margins on their datacenter, and stock went up. This is not enough, as it is just 30% over electricity costs. Maintenance/supervision, and gpu costs/rentals don’t count, and it is unlikely that they are profitable, though it’s not so much accounting fraud as it is accounting PR.

OP’s post is largely right, but it doesn’t require that link to be true. Also, whether these $3m+ systems are warrantied is a relevant question. It’s hard to know exact lifespan from one person saying their gpu failed quickly. Paper still stands well.

Because of power constraints, I’d expect they replace GPUs every 2 years with new generations, and so there will be big write offs.

The AI bubble doesn’t mean AI/LLMs aren’t useful. It means datacenter speculation can’t make money.

those GPUs are headed to the landfill.

They’ll just have a similar discount to the Ethereum switch.

There is a single axis of geopolitics. Pro CIA/Zionazi demonic supremacist corrupt/rigged/controlled democracy evil vs resistors. Tankies are just an insult to the resistors. Speech is controlled by establishment everywhere, and just as money is speech, money/CIA is terrorism that will destabilize and diminish any country that supremacist speech is not exterminated.

This is not pro oligarchist app store. So can only be slavery anti-consumer extortion.

The thing is… side loading becomes more important, and paying less from apps by any other system becomes encouraged. No one likes this shit, but media ignoring actual fascism, doesn’t mean it won’t hit us/them.

Open AI gets so much free PR. Sora is free for now, and instagram getting flooded with copyright infringing crap is not what an expensive video AI creation is going to be paid for. AI videos are very expensive. There are very few people who will pay for it as an alternative for more expensive CGI. Advertising industry can consider full shift for video. They can avoid rights violations and still do it.

Point though, is that copyright controversies are irrelevant to everything important. There is a utility to it, but its not 60gw of power required market.

Video is 14 years old.

do you really think that even when Ai will start (if ever) to generate profits these will be able to repay all the investements done today

First, actual investments that have been done are relatively modest. It’s still a substantial portion of TSMC fab capacity. All of the deal announcements for datacenters are 50x-100x growth. I doubt all of this capacity will be built for a long time. Coding/reasoning models can have more demand, but openAI (most of the deal announcements) is not that good at those. 100x power growth is also 200x every 2 years token output growth, and if models get better, users need less tokens by getting it right on fewer tries.

Second, they are losing money at current levels. Oracle leaked it lost $100m on existing AI datacenter operating losses. Coreweave is fully levered at 10% interest rates. Everyone is operating like social media startups from 10-20 years ago. Only revenue growth and market share, and being cool, matters. Enshittification will come much later.

Third, datacenters are fundamentally flawed, and local AI has competitive advantage to them. AI is good at datamining the datacenter traffic for output that could be profitable to steal.

Fourth, the only business model is US military and disinformation control. They will pay infinitity, and support infinity investment. Giant datacenters are about Skynet. Not market profits. That US government would protect their oligarch partners in stealing your ideas/llm outputs, and amplify current media’s messaging that anti-genocide views are treasonous anti-American sentiment.

how many resources this growth has taken from others places…

If all the money goes towards skynet, energy bills for everyone else will go up, including what little manufacturers there are in US. Insisting on war on China and Russia is helped by forced unemployment, and fascist response to the unemployed’s uppityness. Datacenter AI’s primary certain value is as a new cold war Arms and disinformation race.

Everything is corrupt about this comittee. It needs to be disbanded. All 3 of their nominees were shit.

Tesla stock price has always been valued on “Next Big Thing”. Its robotaxi/fsd is falling flat so far. But robotics is new NBT. With 85% of robots sold in China, and massive manufacturing market share that is still growing, future robot sales will still take place there, and Chinese companies are already well ahead in humanoid and other robotics.

Also their gimmick, robot soccer and robot olympics events, even if not incredibly impressive, is a mass student training program in robotics future that just doesn’t exist in the west. China’s future robot dominance is completely assured because they care about making it happen. Tesla has 0 chance of success, and for a market, requires giving Tesla US monopoly on overpriced robots.

This only got downvotes in another thread. There is far worse that can happen than an AI bubble.

People get distracted over the fate that the pure speculative frenzy could be an AI bubble, and the harm to the hapless speculators and banksters could have a minor impact on the rest of the economy.

Reality is far worse than an AI bubble. It is a US mission for a fossil fueled powered Skynet for Israel that is too big to fail. Bubble in AI investments becomes unlikely, but total destruction of rest of US economy/prosperity becomes assured when the “plebs able to eat in America bubble” bursts is a sacrifice that a fossil fueled powered Skynet for Israel is willing to make.

If Americans are still able to afford to eat, then China or Iran wins.

AMD knows the bubble is going to burst

definitely not it. Anyone making better GPUs should be able to sell them at a price=performance level. LLMs themselves are forever. Running them on consumer level hardware with privacy is an attractive alternative to US military/Skynet allied OpenAI datacenters. That does mean smaller models than “skynet frontier”.

Reaching for other explantions.

CIA allied bankster money to keep OpenAI solvent with some side deal for AMD.

AMD needed a desperate deal to have lead Skynet developer use some of its GPUs instead of only Nvidia. For all the OpenAI committments, there is both massive risk of OpenAI bankruptcy, but also no matter how good other frontier models are/can be, waiting for OpenAI bankruptcy, and buying their datacenters, makes more sense than adding your own 20.5GW (with Oracle) of GPU/power demand, and outbidding OpenAI for GPUs from non-China sources. But its not as though MSFT/GOOG/META can’t outbid for GPUs, or not make better frontier or smaller open models. I could include mechaHitler for Skynet, but the funding pockets though large, seem as though its a marketing/investment headwind for whatever good technical achievements the mechaHitler models have done. MechaHitler AI has its own financing fuckery valuations with Elon share swaps.

Already Mac M3 ultra 512gb is a great LLM machine for very large LLMs (quantized frontier models). Nvidia and AMD will have larger vram consumer GPUs soon enough. Smarter competitors than OpenAI exist in the space, and smartness goes for better cost per benchmark point and per token.

GPT 5 pro is most expensive model in the world at $150/Mtokens output. A $2000 5090 will output the same tokens/cost in 90 days. With privacy or something you can rent to others without Skynet oppression data concerns. On an open small model that you can posttrain to be better at the domain you are interested in than GPT5 behemoth, and rentals that use whatever domain tailored model they want. But US mega corporations already offer 10x

OpenAI losses are accelerating, and expected 2023-2028 total of $45B, with (banskter/OPenAI optimistic) breakeven in 2029. Skynet military contracts of $1T is the prize. Competition to OpenAI can be forced to provide value through smaller model performance boosts that China (Mistral from France too) already actually has the best real world value models. The previous 5090 token output comparison, gets worse on 100x cheaper small open models that exist, but the privacy and posttraining benefits, priceless.

The common weak link in all of these circular financing deals is OpenAI. Except on this one, where AMD is the weaker side. Of course this makes the banksters pump AMD stock. People like to shout AI/LLM bubble, but OpenAI and MechaHitler going for Skynet is going to get a lot of powerful sustain. Still, Skynet ambition is opposite of best path to making money/useful models until US/Israel military payday.

That’s what I meant, though specs look great, and so “unconfident in execution of future roadmap” or cost/yield issues, but tsmc has history of success in new tech generations. OTOH, Rubin is sticking with 3nm as AMD goes for 2nm on the critical part.

Something is very wrong with this deal. Can know this without knowing any specifics of what’s wrong.

NVIDIA’s deal makes much more sense. Will use OpenAI revenue to buy shares in OpenAI and still profit if OpenAI goes bankrupt.

This is less a “vote of confidence” in AMD, because shit is free for OpenAI here. I too will buy all the AMD GPUs if you give me AMD stock of same or greater value. This deal structure actually makes AMD look terrible and unconfident in its product. But deal is not transparent enough.

Dangerous beliefs part of the propaganda that West/Ukraine has been winning all along. Wages are very high in Russia, due to high military pay bonuses, and is massively increasing military production. Russia has enough troops and equipment to be gaining territory every week. Russia moved ahead of Germany and Japan to be 4th largest economy. All of this comes from western reporting.

Just because there is always propaganda about how unconditional surrender of enemy is just mere weeks away, it is suicidal to believe any of it.

While every comment here seems to scream “end patents”, arm has less patent bs than other tech (rounded corners) meant to sue/prevent use. Arm works hard on developing and improving architecture and designs to offer licenses at a compelling price. Qualcomm paying as much as other licensees should be preferable to Qualcomm than bankruptcy.