If You Need a Laptop, Buy It Now | Hacker News Reader

If You Need a Laptop, Buy It Now

theatlantic.com

16 pointsfortran7721 comments

Anyone else expect prices to go down after the AI pop?

Yes, especially if CXMT is able to continue scaling their production and if China is able to crack EUV mass production.

I see RAM prices dropping to new lows in 3-5 years.

Do they need EUV to make RAM? Doing a small amount of searching leads me to 2025 press releases from companies saying they're first to a new process node, and mentioning EUV like it's an innovation.

I assume they could scale faster with more machines of the older, more understood, lithography technology.

In the coming years, EUV mastery will almost certainly be necessary to impact the global RAM market. For context, Samsung started using EUV for DDR4 in 2020. However, DDR4 is generally fine without EUV, it's just as you move into producing the latest DDR5 and HBM3 RAM, the "multi-patterning" techniques that CXMT uses in lieu of EUV result in very low yields (economically uncompetitive). It might be possible to use ultra-low PPP to simply outproduce the rest of the world at competitive prices and maintain reasonable margins. For example, it can cost 10x more in CapEx to build a chemical plant in the USA vs. China. With respect to marginal profits, if China's solar ambitions result in extremely low cost of electricity, their cost for synthetic fused silica could end up being very cheap, and potentially use electrical plasma chemical vapor deposition to create super high quality truly anhydrous silica (if electrical plasma chemical vapor deposition is used). A low price for synthetic silica would also allow CXMT to be competitive at lower yields than EUV achieves.

But over the next 5 years as technology marches on, and high tech RAM production moves from EUV to High-NA EUV to Hyper-NA EUV processes...multi-patterning will likely cease to be feasible. So while China is very much on track to achieve truly astonishing RAM market impact from 2025 to 2030...it then hits a pretty hard wall and China really will need to finish developing and deploy EUV at scale over the next 5 years to achieve market impact past 2030.

I personally wouldn't bet against China achieving a well-scaled EUV deployment by 2030. It's not guaranteed by any means, and reasonable minds might be skeptical because EUV is fucking hard. But they're doing all the necessary things to achieve it: poaching the right engineers from ASML and TSMC, stealing technical documents, funding it generously, and removing red tape. China reported that they got EUV working in pilot plants by the end of 2025, and those reports are considered credible. Another 5 years to polish their EUV tech and scale out EUV deployment seems very reasonable to me.

With how good Claude code and codex are there might not be one.

Yes. However, the economy is also bad due to other factors like unnecessary wars. Things can still get worse outside of the AI bubble.

True, but that leads to even less demand for the oversupply of chips

Where do you envision the pop will come from?

Two possible sources:

1. People who are currently buying AI services realizing it's not all that useful to them and discontinuing their subscriptions. Note that this can come from a changing ecosystem as much as anything to do with the products themselves. I know a couple people running AI propaganda operations where a single person can now do what previously took a major media conglomerate; this is great for them, but if I personally know a couple folks doing this, it indicates that there are probably hundreds of thousands worldwide, and people are simply going to stop trusting anything they read on the Internet.

2. Rising interest rates from the Iran war. Suddenly the cash flows needed to finance all this datacenter and AI model expansion are much higher, and combined with #1, may not be viable.

1. Most AI datacenter plans and valuation are not tied to subscriptions, but from a more vague promise of "AGI," so this isn't likely to pop the bubble IMO (even if it does happen)

2. Historical precedent holds that governments are more likely to suppress rates to spur the economy during wartime.

> Where do you envision the pop will come from?

sudden end of overinvestments in hardware procurements by big players. Its unclear if google for example will sustain 50B/y investments.

Share link: https://www.theatlantic.com/technology/2026/03/laptop-electr...

Technology for things like laptops generally gets better. I still think it's best to buy technology when you need it, not sooner.

Higher-end gaming laptops are still decently priced and work well for local AI inference.

And Linux runs better than ever on them; I'm running debian 13 with almost no driver issues.

For $2k you can get 32 GB DDR5 RAM and 16 GB fast VRAM. Bump the RAM to 64 GB and you're still below $3k.

What models or classes of models would I be able to run on that hardware?

I've asked myself that question while looking at some of the models on this: site https://laptopparts4less.frl/index.php?route=common/home

With 16 GB VRAM one can run a decent quant (Q4-Q8) of newer, smaller dense models. This leaves room for e.g. 32-256k context size.

This might not be enough to chew through a large code base but for smaller projects it can easily fit enough if not all of the code base to drive a good coding agent.

I don't recommend specific models or model providers due to how much hype and BS there is around benchmarks etc. Easiest is to check the latest generation of open models and look for a dense-type where a decent quant fits within the VRAM.

Some models run fast enough that some of the weights can spill over from VRAM to RAM while maintaining a usable prompt/token gen speed.

Thank you!

https://archive.ph/FoWIt

Everyone panic.

Will the Oracle layoffs flood renewed ThinkPads on the market?

Holy crap lol

EDIT: Oracle will prolly just destroy them. Most companies do.

Reply on news.ycombinator.com