Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The US is winning the AI race where it matters most: commercialization

avkcode.github.io

Post reply on HN
Print view View original

311–320 of 718 posts

Re: The US is winning the AI race where it matters most: commercialization

#311

Post by visarga »

aspenmartin wrote: post #298
mordae wrote: post #277

No, they are not. They are winning because West is forbidden to use Chinese models for anything work-related.

You’re saying if we were allowed to use e.g. qwen more broadly the US wouldn’t be in the same strategic position? We have the best models…we own all the companies that make the best infra and the hyper scalers…I don’t think “oh we can use Qwen now?” Would exactly devastate the US

> I don’t think “oh we can use Qwen now?” Would exactly devastate the US

You'd be surprised how useful it can be to fine tune it in enterprise.

Re: The US is winning the AI race where it matters most: commercialization

#312

Post by akrylov »

jmyeet wrote: post #93

Back in the dot-com bubble, people started inventing new metrics to "value" dot-com companies that lost money hand over fist. My favorite was "revenue multiples". So instead of a a P/E ratio, it was just a multiple of revenue no matter how much money you lost. We've invented a new term here too: revenue backlog. OpenAI and Anthropic in particular need to recover probably at least $2 trillion to recoup their capex inv…

It's not "AI bubble" - at this point it's a software bubble. It's Antropic or OpenAI that should justify their valuations, they have close to billion customers at this point. It's non-AI software companies without strong cloud business that must justify why their core product is not going to be replaced with the help of LLMs. It's not "fair" but that is how it's seen by the Wall Street.

Re: The US is winning the AI race where it matters most: commercialization

#313

Post by CMay »

watwut wrote: post #258
CMay wrote: post #186

Earlier quoted context omitted.

The US is not xenophobic. That is ridiculous. Any time you say stuff like that, you discredit the things you say that actually make sense. I'm with you on the privacy aspect, but there are multiple dimensions of that which you're ignoring. I'd much prefer taking my chances in the US than in the EU, where they are constantly trying to push companies to weaken privacy.

Europe seems both better capable of sustaining democracy, privacy and rule of law. USA is on verge of being irreversibly done for in all three areas. It can happen in Europe too, but the full fall is not that close.

Privacy is a concern everywhere, but the center of gravity of the issue moves further up or down the chain depending on the country.

The structure of the US makes it basically the single most secure democracy anywhere right now or in history. No country in Europe or Europe as a whole is even competitive by comparison. The main issue we're facing is that we are by far the primary target for foreign funded activism and systemic attacks, because China and Russia hated NGOs promoting color revolutions.

That is also part of the rule of law issue, but the system is overall managing quite well. It's all moving in slow motion, but many important metrics are going in the right direction, which we need as that's part of deterring China.

Re: The US is winning the AI race where it matters most: commercialization

#315

Post by Havoc »

jasondigitized wrote: post #280
Havoc wrote: post #169

Inclined to disagree. The winner here will be whoever can move atoms with AI not take notes at the daily standup. i.e. Think boston dynamics vs unitree They're both doing well but I'd lean towards China is winning on atoms in light of a huge manufacturing base they can AI-ify.

The winner is whoever can move the atoms for free, e.g. crack energy.

Mostly agree. I think there is a big time delay though.

If free cheap energy is unlocked today I reckon it would still take a good 30 years for that to ripple through properly.

It solves lots of problems (water!) but doesn't make the heavy machinery to consume it instantly appear.

Re: The US is winning the AI race where it matters most: commercialization

#316

Post by ripvanwinkle »

This para caught my eye

>Frontier cyber models may push states and defense firms toward the opposite logic: security by obscurity, with closed software, closed tooling, closed firmware, and closed chips. If a model cannot train on the code and architecture of a target stack, it will usually have less context and less speed. That does not make systems safe, but it does raise the value of proprietary stacks all the way down to hardware.

Is this really true. Are there any experts who can weigh in on this.

Should we interpret this to mean that in the new world Windows is more resistant to attacks than say Linux.

Re: The US is winning the AI race where it matters most: commercialization

#317

Post by kelseyfrog »

Jalad wrote: post #135
rich_sasha wrote: post #65

Earlier quoted context omitted.

It's a war in the sense that there's a concern that eventually you hit a singularity and can outsmart others in ways not constrained by human scales. If you make better guns, you're still limited by how many people can carry them. You can't conquer the world just like this. But if someone invents super intelligence, they can dominate new AI research, control global economies, fight much better, and all very quickly.

> But if someone invents super intelligence, they can dominate new AI research, control global economies, fight much better, and all very quickly. After reading "If Anyone Builds It, Everyone Dies" I think this is not the correct take. If anyone creates ASI, it just means it's going to wipe everyone out, and it doesn't matter if China or the US do it first

If you were an American, wouldn't you prefer the US wiped you out rather than China?

Re: The US is winning the AI race where it matters most: commercialization

#318

Post by ipython »

ripvanwinkle wrote: post #316

This para caught my eye >Frontier cyber models may push states and defense firms toward the opposite logic: security by obscurity, with closed software, closed tooling, closed firmware, and closed chips. If a model cannot train on the code and architecture of a target stack, it will usually have less context and less speed. That does not make systems safe, but it does raise the value of proprietary stacks all the way…

I think there’s some credence to the concept that more context == faster iteration cycles. Source code can be one major source of context.

I think “security through obscurity is no security” concept was aimed toward people not relying on obscurity alone as a security mechanism. And largely that message succeeded. But now we are in a rapid acceleration of capabilities (on both sides) where any advantage to one side will result in outsized gains, at least in the short term.

Re: The US is winning the AI race where it matters most: commercialization

#319

Post by visarga »

throwaway27448 wrote: post #86
SubiculumCode wrote: post #38

Earlier quoted context omitted.

That seems like a lot of rationalization to me. China is pursuing these because they cannot compete on the frontier. Yes, there is a possibility that all that compute is not needed, but it is a rather remote possibility, and there is no doubt that, given the choice, China would be pursuing frontier model building with closed, propietary-only offerings.

> China is pursuing these because they cannot compete on the frontier. ? Claude, ChatGPT, etc are heinously expensive for tiny benefits lmao. Local + efficient is clearly the future

> ? Claude, ChatGPT, etc are heinously expensive for tiny benefits lmao

Unfortunately local inference is inefficient, 100s of times more inefficient than cloud. When you answer one request at a time you still have to fetch all active weights into compute units, once every token. When you run a batch of 300, you load it once and compute 300 at a time.

Compared to cloud, local inference is less flexible. You can't scale up 5x or 20x, can't have spikes, and pay for it no matter if you use it or not. But usage factor is very low, like 5%. And to run a decent model your system costs $2000 or more.

Re: The US is winning the AI race where it matters most: commercialization

#320

Post by wouldbecouldbe »

eskibars wrote: post #309
jxf wrote: post #286

Earlier quoted context omitted.

My understanding is that it's not that the _models_ are banned, but rather the _platform_ is banned. It is acceptable to host, say, `deepseek-r1-distill-qwen-7b` and run it yourself, for example. It is not acceptable (to the authors of these bans) to download the DeepSeek app and run it on your work device.

I just left a job for a German B2B software company which sold primarily to large automotive, defense, and aerospace companies. Several of our customers specifically banned anything with the word "DeepSeek" -- hosted or self-hosted. There's still a lot of naivety on what the difference is between models and platforms, and its easier for a lot of these big companies to just make a blanket statement like "nothing DeepS…

You are sending all your prompts code and files there. So ofcourse its an issue
Post reply on HN