> I've skimmed the thread here and I am now seriously considering leaving HN for the first time in about 15 years.
I'm finding a lot of the comments here pretty reprehensible, but no more reprehensible than the collective shrug the community gave towards murdered Palestinians, or threads about dead Iranians as a result of American bombs that get flagged off the front page. That doesn't make them acceptable or okay.
Those people's lives are/were valuable, too. It's disgusting that we try to keep HN "clean" of those horrors and the people that flag those threads should be ashamed. Ditto those who think the killing of innocent civilians is okay.
Well, you know, dead palestinians aren't paying their salaries or investing in their companies, so they aren't as important as a accelerator that in the last batch had 90+% of 'AI' companies.
Think of the investments they may lose. We can't have any of that can we?
> If we want men to take up certain roles, we need to pay more.
Why is it only now, when employment rates are seemingly a problem for men, that we need to pay more in these professions to attract men that might otherwise not have considered them? Why shouldn't we have paid more earlier?
The framing of the article and discussion around it is a little bizarre to me because it ignores the decades or longer of (American) society effectively pushing women into industries like education or nursing and subsequently devaluing them.
I don't quite understand why society has to step in and try to fix this for men who are feeling insecure about their job options while simultaneously actively avoiding trying to help women and minorities.
It's funny because arguably a big reason Harris lost was unhappiness with the cost of living/the vibecession. Yet Trump hasn't managed to fulfill his promise of turning the magic grocery price knob down (they still seem quite expensive to me) and I'm still skeptical there will be any kind of "blue wave" come November. We'll see if voter sensitivity to living costs only apply to one party or not.
Hasn't seemed to catch up with Tesla, which is still highly valued despite making a pretty mediocre car compared to the competition. Even if one makes the argument that Tesla vehicles are of good quality, it's still a high valuation that seems to show no sign of dropping.
> Bottom line is that H100 prices are near 3 year highs, A100s are still profitable to run, B200 prices are increasing, no one has enough compute.
Then why aren't the hardware manufacturers of components needed by AI companies making plans yesterday to bring new fabs online to meet demand? That isn't a gotcha question, I genuinely want to know. The money involved isn't that much compared to the money changing hands between Nvidia Microsoft, OpenAI, etc., and it's not like once in-progress data center construction is complete they won't need to buy more RAM and GPUs, especially with any new advances in technology that might happen.
Inevitably someone will reply that hardware manufacturers don't want to be stuck losing money on a facility because the bubble popped and demand disappeared, but if Anthropic and OpenAI are going to "run laps around current big tech", it should be a no-brainer to increase production capacity.
A new fab will need to be filled with advanced equipment like lithography machines. They are the most complex thing humanity has every built.
There is one supplier of EUV lithography machines in the world, ASML. They are basically acting as an integrator for hundreds of highly specialized components manufactured to unimaginable levels of precision. Each of them has roughly one eligible supplier in the world who are operating at full capacity. To expand, they'll need yet another set of specialized and almost impossible to build equipment.
So the supply chain moves incredibly slowly, and the slowness is intrinsic due to the complexity and depth of the supply chain. It can't be fixed with just money. IIRC ASML is aiming to merely double their production of EUV lithography machines by 2030.
Sure, I didn't mean to suggest that it would be easy or fast to increase manufacturing capabilities, just that the confidence I'm seeing around AI should extend to the manufacturers (if that confidence for the future growth and success of OpenAI and Anthropic is warranted). That is, the business decision to increase RAM and GPU supply should be "easy".
They are. They're making as many fabs as they can as fast as they can.
The bottleneck is ASML, who can only make so many EUV machines. No one else can make EUV machines.
Scaling chip fabs and chip equipment is much harder. And you have to understand that chip fabs go bankrupt if demand suddenly drops so they have to be more cautious by default.
If you're really compute constrained do you really need EUV machines? You can make do with DUV fabrication nodes, albeit at somewhat higher cost. The trailing edge is where a lot of the mass impactful innovation is, e.g. trying to replicate more advanced EUV nodes with DUV multiple patterning.
That’s what’s happening. Companies who were planning a move to advanced nodes for non AI chips are delaying it. All the advanced nodes are going to AI or smartphone chips only.
There was a good episode on Dwarkesh's podcast about this in the last few weeks, just a deep dive into the semiconductor industry and what the bottlenecks are.
They used to, but this generation has been hit by at least 4 independant crises that have made them more expensive over time:
1. Moore's law dying means that older nodes are still very useful, so there's still lots of people bidding for capacity on these 8 or whatever year old foundry nodes who are willing to pay lots of money, meaning that base costs of the CPU and GPU haven't fallen as quickly as one would expect even in the absence of 2-4.
2. The Pandemic and then Russia's invasion of Ukraine screwed with supply chains quite a bit, and caused an inflationary spike.
3. Trump's tariffs affected the profitability of these consoles, and inserted a lot of uncertainty into them because nobody knows how the tariffs situation will evolve over time, or if Sony will get reimbursed or not for the illegal tariffs that were levied against them. The uncertainty and general animosity towards the USA has also caused the US Dollar to slump in value relative to a lot of currencies, which then pressures Sony to raise prices.
4. The current RAM, SSD, and GPU shortages caused by LLM hyperscalers is again spiking the costs of their components
I'm also very skeptical of "everything eventually passes" as it pertains to hardware prices. Right now, prices are high because supply can't keep up with demand. But if/when supply increases to meet demand or demand decreases, there's no reason for companies to drop prices now that consumers have become accustomed to them.
Exactly. Production of RAM, SSDs, etc is spread out enough that no one company/country/fab has a stranglehold on the market. Right now anyone with a memory fab has a money printer. More people will build fabs, just like they did last time. It takes a bit but they'll get built.
> I was under the impression that only ASML could build SOTA fabs.
You're correct that this is true, but wrong in that it is not relevant.
SOTA fabs are used for cutting edge CPU/GPU chips. You do not need a SOTA fab to create DDR5 memory or SSD memory. There are a number of companies able to create the lithography machines to do so, and dozens of actual operators of those machines.
Why would that concern you unless you are working on the cutting edge and the very limits of that hardware?
The current generation is insanely fast. I am planning to get a gaming PC for my wife and a mix of gaming + workstation PC for me (or maybe just base it off of the Ryzen 9950x3D and call it a day). We plan to hold on to them for 10 years.
I don't care if anything 6x faster comes out. For what I need the current generation is even an overkill.
I'd even go as far as to say that it would be quite OK if that's the very last generation and no further hardware development ever happens.
I am on the edge of current available hardware and do feel the desire to upgrade. As stated before, I am unhappy with the current maximum when combing frame generation, resolution, and graphics quality.
My dream spec is UE5 at 120hz on an 8k oled. I think that sounds like a super sick experience I would buy tens of thousands of dollars of hardware for.
What's boring to me is how abstract many of the "AI success stories" tend to be, even on here. A whole blog post about some new way to use LLMs, or a best practice, or whatever, and no link to the code or dotfiles. I understand that how you prompt is a big part of things but all the major providers have a lot of configuration options. There are whole ecosystems of plugins.
It's just not very interesting or useful to me to read about how you got AI to output better quality code or how you can program from your phone now without going into detail. And so many of the conversations are showing off the wins without talking about the tools, configurations, or other parts of the setup that made it possible.
It's still around and up to 6th edition! Catalyst Game hasn't been the best steward of the IP, with the rules still being internally inconsistent and usually needing a lot of house rules to fix.
I'm finding a lot of the comments here pretty reprehensible, but no more reprehensible than the collective shrug the community gave towards murdered Palestinians, or threads about dead Iranians as a result of American bombs that get flagged off the front page. That doesn't make them acceptable or okay.
Those people's lives are/were valuable, too. It's disgusting that we try to keep HN "clean" of those horrors and the people that flag those threads should be ashamed. Ditto those who think the killing of innocent civilians is okay.
reply