maxglute

Chinese peasants been starving for 1000s of years, it took Chinese communist speedrunning development at the expense of human lives a few decades to end that cycle. Same with Russian serfs. There's a reason PRC lifespan and industrialization was significantly higher than developing peers by the 70s, well before market reforms, fail fast, break eggs make omelette etc etc.

These were ~Q1-Q2 commitments based on what Anthropic THOUGHT token maxed compute demand was going to be over next few years. Then Late Q2, and Q3 PRC models dropped floor for token prices. It may very well be that Jevons demand which Anthropic though required 500B of compute now only requires 100B, and unless demand is 5x what they thought, those additional 400B non hardware cancellable commitment are going to be balance sheet concrete shoes.

They're fine, I'd rather have zelda with modding.

Would be lovely if hardware pricing and AI finally killed consoles and exclusives. Not that expensive hardware price wouldn't also fuck pc hard, but fingers crossed some time cone where where things peg around a common PC config like steam machine. I know the more likely path is locking big budget AAA games behind streaming.

North Korea would two birds one stone pretty hard, fucking with Japan and gain all the soft power if they spend some of that hacking talent pool fulltime Sony and Nintendo emulation... using American AI.

Which of my words implied any of this?

One person's hobby started the industry in PRC and set the trend and inspired talent in nascent field with no established business model. PRC competitors didn't want to FOMO and just knew they had to compete with opensource and everyone fast followed. They bought what limited compute they could, hired what talent they could and just told them get to work. There's a reason even Chinese Uber / DiDi was releasing open source models. There wasn't some grand coordinated posture to compromise US markets, it's incidental to PRC industry being PRC industry. Very few in PRC thought they could close gap this fast with compute deficit. US chose to continuously dump venture $$$ and company warchests quarter after quarter on their own to increase exposure. There's no grand conspiracy to turn US into communist state (wut), there is not "bet", just typical PRC domestic involution business shenanigans clashing with US financialized VC shenanigans. In alternate world if PRC models not competitive, open source wouldn't even matter to US valuations. In alternate world if PRC models competitive and US AI isn't leveraged to shit to multiple % of gpd, i.e. systemic contagion risk, it also wouldn't matter.

It doesn't matter if US/west makes $ off PRC models, PRC doesn't have enough indigenous compute let alone spare to meaningfully compete globally. In so much it matters economically (again happy happenstance for PRC) is denying US AI making $$$$$ while spending $$$$ vs US AI making $ while spending $$$$. Extra happy happenstance that $$$$ US spending on is basically all the VC money and FAANG warchest stockpiled by US tech over decades... i.e. US spent so much that it meaningfully constrains private money to spend more in future. Meanwhile PRC AI spending $ to make $ (or less), but still have $$$ in reserve for other shit. It also doesn't help that half of business model depends on building infra on tight schedule that US is... not great at.

This really not that hard.

Deepseek started as Liang's side project funded with his hedgefund couch change compute. That set the economics for everyone else in PRC. If deepseek releases for free because Liang doesn't give a shit about making $$$ on his hobby, every other PRC player has to follow. And serendipitously, due to US export controls, PRC tech companies can't dig themselves into ruinous seppuku debt buying compute even if they wanted to - what they can buy, i.e. everything around DC infra including power, they get for fraction of the cost expediently.

TLDR Liang said fuck profits... which means every other PRC competitor had to fuck profits, which ends up fucking profits of western labs so deep in debt due to rampant financialization and can not afford to say fuck profits.

Sometimes a billionaire is fine just burning $$$ on a hobby while remaining a billionaire instead of trillionaire, and it just so happens the side effect might mean preventing others from being trillionaires as well. Bonus others is geopolitical adversary who is hedging economy on trillion $$$ returns.

I mean there's a lot of side quest slop in best Ubisoft games, agents generating better content in future isn't our of question for the development pipeline at which point clone me a reasonable assassin creed 15 with roughy xyz story might be good enough. Shared mainstream tv culture died with streaming fragmentation, people playing different versions of games which stop being canonical is not unforeseeable.

Cope never ceases.

Reminder heavy REE i.e. the strategic stuff that goes into sensors (what modern US MIC is built around) is geologically rare - economically extracted from limited ionic clay deposits, which is much in fact rarer than... making magnets by smashing rocks for light ree.

That and the entire HREE economic extraction pipeline is Chinese, who build out the foundation tech over last 30+ years aka capability US/west has ever had. The physical reality = agglomerated Chinese mining cities of millions with vertically integrated industrial chains sustained tertiary mining/engineering pipeline larger than RoW combined.

Let's be real this not scale US capable of replicating even if it had ionic clays, and even if US manage to sort out LREE (mountain pass etc)... it still basically built its entire strategic MIC (i.e. sensors) on HREE that only PRC can process at scale. Which is kind of dumb, but here we are. But that's not going away for a long time, i.e. PRC will likely scale EUV to western levels well before US/west can scale HREE to Chinese levels, maybe F35s getting real radars before that.

Another nafoid shibboleth/iq test.

If one believes Zitron / Microsoft claims that like ~50% of AI chips bought are in warehouses, and current projected DC delays... the TLDR is the amount of compute lead (aka what drives lead) is fraction of projection, which will shift timelines accordingly. And really if DC build out delays is due to infra or scrubbed altogether, we're talking about years.

Startup spending a few million dollars to decompile and replicate adobe suite in a few years would spring up so many competitors that it would basically kill adobe and creative software sub. That's at least a more useful allocation of tokens than logn -182. Porbably prosocial movement.

It's been 20 years, open source more usable for typical use case but gap/ecosystem expanding.

Honestly, I'd be surprised the Adobe or current workflow even survives in a few years with people who grew up on image prompting and sketch to image etc.

But it'd be nice to someone to throw enough tokens to make some parity Adobe software... like OpenAI wasting 10m on 1m math problem poor waste of resources... spend 100m and kill and charge $10 instead of adobe $50.

TBH only scanned through ~20, impressive technically but disservice to Invisible Cities to render it as... uninspired, bland, miniatures. All the wrong of details to focus on to capture even notional urban verisimilitude. Looks like a 8 year old drew scribbles and their dad finished the project.

Nortelstans: You took everything from me.

Huawei: I don't even know who you are.

Terminally stupid cannucks still butt and misattributing Nortel primary fell to HW stole muh secrets, not that Nortel grossly mismanaged (and outright fraud). Hint PRC stole lots of shit back then... but Nortel fucking exploded spectacularly. Why didn't CISCO explode because HW copied some homework... because unlike CISCO, Nortel was mismanaged to shit. NVM HW wasn't even state champion at the time, that was ZTE, they didn't juicy state level industrial espionage perks. Meanwhile allegations against HW vs Nortel was over optical routing, i.e. niche shit to HW overall portfolio. It's just Canadian cope.

This not mentioning Nortel basically brought this on themselves by bamboo ceiling the shit out of Chinese employees in PRC (both international/PRC nationals). Nortel HQ/Ottawa -> North PRC/Guangzhou, i.e. from Canadians to Chinese Canadians basically all defected Nortel (and poached entire R&D teams), because North PRC wasn't giving them their dues... and not just to PRC SOEs, but western competitors like lucent/acatel etc. Defectors were bitching / organizing in my local Counter Strike server, their AFKs very disruptive, so in a sense I was just as much victim as Nortel.

US domestic law for US territories, AKA it's Super Ineffective!

Huawei (with PRC support) will sue/sales injunction international OEMs/manufactures that uses QC... for consumer electronics, QC doesn't make shit, US barely make shit, hence QC PARTICULARLLY fucked because essentially every producer of QC in every segment goes through PRC supply chains, including US partners like South Korea who manufactures Vietnam but depends on PRC immediate goods. PRC tells them to stop making shit with QC, they stop or PRC makes sure they stop making all shit. If device makers can't use QC, they use something else -> QC is kill.

And unlike domestic US laws, SEP enforcement legitimate/international, i.e this not where US entity lists and tries others to not HW buy, QC just becomes internationally "entity listed" for violating SEP.

There's a reason US/BIS carved exception immediately. Because this not lawfare battle US can win. And vice versa, HW still paying for US IP post sanctions and pays when losing international IP battles, i.e. 2B to QC of all people... over SEP in 2020. Because it's better to not fuck with SEPs if you want to sell globally.

But really take to logical conclusion, US global market/manufacture exposure existentially higher than PRC, US has more markets to lose and loses them harder. PRC can bypass US and make shit, US can't vice versa.

Charitable interpretation is Huawei actually had R&D + academic linkages in Canada (was expanding aggressively globally), but Canada still drop in bucket, Canada R&D headcount peaked ~1k vs ~100k in PRC. Post sanction, Canada basically siloed from doing crown jewel work anyway.

IIRC US/BIS carved out exemptions for intangibles like SEP patent licenses. Entity list for controlled goods export/import. Regardless, US would also be stupid not to pay Huawei, PRC pays $4 for every $1 she receives for IP, ~30B deficit, most to US. PRC takes home more $$$ if US tries to lawfare it's way out of paying. But bigger stick is PRC entrenched in supply chains enough that Huawei (with PRC pressure) can sue OEMs (patent enforcement happens at device level)... all domestic PRC manufactures, foxconn, samsung, apple (really anyone that uses PRC supply chains at all) on chopping block, i.e. functionally global sales injunction against any 3/4/5g hardware that uses QC which has to use HW SEPs, which covers their entire portfolio... mobile, auto, iot, edge comptue... aka QC dead if it doesn't pay.

How the turntables etc.

Pre sanctions, HW fine buying QC for mobile chipset StandardEssentialPatentswhile QC pay HW for 5g SEPs, with net balance favoring QC. Huawei indigenized mobile SoC + 5g hardware with Kirin 9000 in 2023. Hilariously, now that Huawei weaned off QC SoC... QC simply stuck net paying Huawei.

3. more accurate framing: replace dark world thinking with realism and frame liberal middle class western bubble with libtard delulu.

4. it's 2026, the sentiment have shifted from PRC insecure to PRC IS strong. Apart from national soccer eon of humiliation - a devils bargain to transfer all national failure into soccer, I think last 10 years has proven PRC/chinaman working twice as hard has paid off across domains. Chip on shoulder still remains, of course there's misguided netizen comment on foreigners in general, but even post china-maxxing general foreign perspective on PRC across domains is every bit as myopic and ridiculous.

Like a fancy dac. The specs are very pedestrian for a music device.

Looks massive and 20hours of battery life. Does 3.5 even have a dac?

Is there reason antigravity has 3.8, 3.7, 3.6, 3.1 and old ass claude / gpt models in the drop down. like why is this not streamlined or deprecaed models removed.

TFW Jevons arrives from supply side not demand side.

Western labs: Not like that!

Why always CCP this, CCP that. The parsimonious answer is this shit started when Deepseek founder decided to open source, because at the time the Whale was just hedgefund side project that serendipitously set the tone before they were even on CCP radar. Then other PRC domestic players have to copy / involute to free to be competitive, but due to sanctions / they are compute constrained and can "afford" to, because sections ironically means PRC AI can never dig themselves in the bottomless debt pit of western labs.

TLDR of timeline/chain of events: some billionaire hedgefund manager with AI as hobby, said fuck economics, let's opensource AI... and all the other players have to now compete with that market force. Of course it does not mean equilibrium will last forever, we already seeing defections from model, but imo thats basically what happened... singular hedgefund bro said fuck profits... other PRC players also had to fuck profits, which ends up tripping western labs so deep in debt who can not say fuck profits.

I am not saying policy makers have not been responding, I'm saying it's not existential urgent to bring out never before punitive natal tools. It's still not pressing problem for decades, now is feeling out phase, i.e. doing simple things like dropping policies or trying to clean up childcare involution.

> started dropping

This presumes PRC wants to maintain 1.4b baseline - imo it doesn't. Currently only ~800m people are actually productive. Initial decades of population drop will be disproportionately shedding bottom quantiles of old / uneducated / could not be incorporated into modern economy. Bluntly they're useless bodies - there's a reason PRC has 200m farmers... when 20m +machines will do. Or 100m other make work jobs for idle aunties and uncles. If PRC not aiming to maintain 1.4b... frankly considering mainland resource stress per capita they shouldn't, then it's fine and preferable to bleed pop for a while.

> reduction in economic demand.

No, this orthodox economic idiocy. The bottom 2 quantile constitutes like 5-10% of GDP. They are/will never big sources of demand to be begin with. More than offset by generating 100m productive advanced workers who has 10-20 times the income/spending power. Of course not all demand same, but in aggregate better to trade a 4 rural grand parents living on pittance of governance transfer and savings for tier1-2 stem chad who will add more value and out spend lifelong vs turnip eaters. Cohorts are not economically equal, sometimes grossly so. Of course you can't simply liquidate them, but doesn't mean governance will miss them when they expire.

> years from starting to contract

Consider 2 framings:

1. orthodox: PRC workforce decrease by 30%. Line must go down. 2. reality: PRC workforce optimizes to be 100% more productive with 30% less head count, by trimming all the fake work jobs doubling their their productive staff. Line goes up.

Again cohort size less important than cohort profile/productivity/composition etc. On way out - mostly old, unskilled, farmers, informal workers - the who gamut of make work economy that nets single digit percent of GDP that gov BEGRUDINGLY supports because people need jobs - it does not mean state/economy needs those people. Many of these are jobs that policy makers wanted to automate away for years but could't, like PRC agri literally being held hostage for years to cater to small plot armer inefficiency.

The only real labour shortage is skilled bluecollar/vocational, and that's mostly because new gen Chinese kids are too proud to work dirty jobs, hence contradictory youth unemployment problem. There's plenty of domestic bodies to staff workshop and worksites. PRC just in mass tertiary transition cultural adjustment period where grad would rather do flexible gig economy while looking for god job than demean themselves with peasant work. This new phenomenon, scaled to tertiary explosion timelines, they'll eventually give up / come around.

Now of course long term demographic / fertility management important and pop cannot crash too hard. But understand PRC now, all the involution, costs, delay to family formation is because they front loaded talent generation so hard that there is fancy degrees + air conditioned jobs mismatch. In 20-30 years structural forces that converge is pop down, housing cheap from oversupply, wealth transfers, signs of future labor shortage... i.e. all signals that stack up to say, if you have kid now, they going to be cheap to raise and guaranteed job in future by virtue of no competition. Basically inverse of current decision making.

Again that just shifts curve - does not get to replacement, like ultimately countries that don't want mass immigration have to figure out other inducements. But that's too in future to usefully project, i.e. we could be spamming test tube babies in state orphanages.

Fertility really 20/30+ year away problem.

PRC growth now based on educated population, which exploded, not total population. Reality is every tertiary median to upper middle class wage is more productive than 10 undereducated peasants... the ones first to age out. There's like 200-400m+ bluntly useless mouths dragging down productivity/per capital metrics, the only saving grace they did NOT get rich before old, i.e. poor old so it's cheap(ish) in terms of resources (but not labour) to support them. Put it this way, PRC is generating OECD combined in tertiary, disproportionately STEM, they have backlog of 200m births from last 20+ years that still haven't got their masters/phds... at 60% and increasing tertiary we're going 100m+ high end talent, aka the greatest high end human dividend in recorded history who will be in workforce for until 2060s, i.e. so far out its hard to project based on present realities, only they have no shortage of advanced economy growth potential. Like if you throw all projected US newborns + immigration into STEM... PRC would still come out FAR ahead. There is no contest, unless AGI displaces everyone.

TFR real issue probably 30 year out. There will be some rebound, current crashing TFR because so many entering tertiary last few years = delay family formation. Women going to turn 35-40 and realize they want kids... and have them, especially considering 1:4 dependency ratio also = 4:1 wealth transfer + property going other way when previous generation (with high savings rate) dies. But won't get to replacement level. Then it's matter of what bad levers PRC willing to pull - pro natal fertility policies does not incentivize 2.1 TFR... and no one has tried punitive policies, i.e. limit wealth/property transfers to families with 2 kids... all sorts of inducements liberal societies haven't... or politically can't try. Or immigration, test tube babies, who knows. But IMO they try punitive policies first. That's really the only thing that no one has tried, and we know from wealthy Muslim countries with oil money, religious incentive, domestic help... aka every support imaginable, TFR is below 2 or trending towards it.

But why? Someone got a woody after seeing topgun darkstar (fair). But why.

If Uber caps employee AI spend at $1500, then they've just admitted per employee AI spend on par with premium health care or commercial rent both in the 1000-1500 range, and this limited to just technical employees, i.e. subsect so presumably AI as % of business cost even lower. Health insurance 1T industry... commercial RE like 100 billion...

Anthropic thinks it's worth more than insuring all US workers AND putting roof over their heads. AI sector thinks its worth magnitude more in aggregate. If they fully displace people, i.e. untied to headcount eventually then sure, but if not, based on what Uber pays, AI is basically worth as much as insurance.

He literally sets / signs off on industrial policy, others follow, This not malaria/polio/smallpox where one can attribute credit and proliferation to X team, and Y policy. PRC aggregate renewable spans 100s of sectors and programs etc, only sensible for credit to flow up to the big man. If one wants to split hairs, can also credit Hu who did indigenous innovation and renewable act, but execution and magnitude increase in industrial capacity expansion realized under Xi. Like literally no one else has stamped and enable industrial policy across sectors to displace current/future emission on order of magnitude of PRC renewable. World power demand is going to increase magnitude via simple 50% percent of world developing... MIC2025 essentially single handedly shifting future curve towards renewable from fossil. Teslas will never displace more emissions than whole of PRC renewable/ev etc. Elon simply is not in the running for global scale renewable adoption and emission displacement. No one is.

> possible that their CO2 output will increase.

Immaterial if CO2 increases, only useful geopolitical metric is historic per capita emission relative to development level and PRC is low for country blasting through capex heavy capital accumulation phase, i.e. PRC setting standard for low relative per capita emission development. Xi 2012-2026 did ~30% global emissions, ~10-15% global historic for ~20% of population. As long as PRC stays under 20% global historic and maintains above 50th percentile development (use whatever proxy indicator, per capital income PPP etc), they are under emitting / developing efficiently relative to energy use. AKA PRC developing more for less unit of emission. The alternative is not developing which frankly stupid argument/position. For reference US/Europe ~45% historic emission for 10% of global pop. On balance PRC currently grossly under emitting vs historic developmental baseline.