|
It’s The Math – How China Is Avoiding The AI Bubble
The Artificial Intelligence bubble in the United States is still growing. It is consuming a large share of the available financial resources. But its output, in form of real products, is so far rather meager.
There are some code generation tools which are, at times, helpful, but, unless subsidized, very expensive. There is ChatGTP and other squawk boxes which in the end are just new forms of inherently unreliable web search tools.
What is missing are useful mass applications billions of people are willing to pay for.
Still – a large number of ‘very important people’ believe that the Large Language Models, which are at the core of OpenAI’s and Anthropic products, will one day reach the capabilities of sentient beings. That is, in my view, utter bullshit, but who am I to tell you.
The NY Times has a long write up (archived) about Oracle founder Larry Ellison and his bet of nearly everything he owns on the A.I. bubble. It states:
The story of A.I. has been as much a financial story as a technological one, a question of how to structure the mind-boggling investments required to train and run the models. Few people doubt that this technology is going to change everything. What’s less clear is when the profits are going to start rolling in and how big they are going to be. “To me, it’s a math problem,” says Asad Ramzanali, the director of A.I. at a policy center at Vanderbilt University. “We are making trillions of dollars in investments on the back of tens of billions of dollars in revenues.”
That is why Ed Zitron and others will rightly tell you that the huge investments spent on AI are making no sense at all.
Where are the products and the customers that will allow to recuperate the hundreds of billions of shady dollars spent on data centers for AI?
The financial structure of the data center build-out makes it especially vulnerable to a crash. The deals themselves are built on enormously complicated debt and equity schemes that involve circular financing. The hyperscalers are investing heavily in the same companies they are counting on to buy their computing power. It’s what economists call an interlocking liability structure. If their customers struggle to monetize their products, they will be hit extra hard — and so will their investors, which include a lot of everyday Americans. And these are just the U.S. companies. The A.I. boom has been a global phenomenon; an A.I. collapse would be as well.
The NY Times presume that everything the U.S. does is copied throughout the world. If the bubble burst in the U.S. it will, in consequence, also burst in other places – especially in China.
This is a misunderstanding of what the Chinese AI models are, and what Chinese AI companies are doing.
Throughout the last year several Chinese companies have surprised the public by offering AI models that are nearly as capable as the best ones U.S. companies produce, but are offered at a price that just a tenth of the cost of the (heavily subsidized) U.S. models.
Moreover Chinese companies have published the weights and source code of their models and allow anyone with the adequate hardware to run them on their own premises.
The Economist has made an attempt to understand this:
How China gets better bang for its buck than America in AI (archived)
Its investment lags far behind America’s. Its models do not
The American AI splurge looks especially profligate compared with Chinese parsimony. In 2026 Chinese tech titans are forecast to invest less than a tenth as much in data centres as their American counterparts (see chart).
Their models appear only fractionally less powerful owing to this frugality. K3, an advanced model launched last month by Moonshot AI, a Beijing-based startup, is 95% as clever (on widely used benchmarks) as Fable 5, a frontier model from Anthropic, another top American lab. It is also 70% cheaper to use. On August 3rd Alibaba, a Chinese tech giant, released a model which reportedly scored among the world’s best by some measures.
The authors at the Economist go on to muse about cheaper data center costs in China. The country, they say, has also less chances to invest in overpriced U.S. hardware. It’s buyers are more stingy and the search for real applications is more important for China than the quest for a ‘singularity’ or some ‘general artificial intelligence’, which U.S. companies pursue.
This is all, like in the NY Times piece quote, mostly bullshit.
The real discriminator between U.S. and Chinese AI companies is the math they use.
U.S. models run on algorithms that are computationally wasteful, while Chinese models use smarter methods, which need less compute, to achieve a similar quality.
Language models, which, given some text input, produce a related text output, are nothing new.
Most are based on so called neural network algorithms which are a simulation of a rudimentary function of the human brain. Neural networks can be trained to recognize an input to then produce a related output.
The problem with neural networks and their early variants is that the training is very time consuming which limits the size of the models.
In 2018 researchers at Google found an elegant way to circumvent these restriction:
The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train.
U.S. companies jumped onto the Transformer model. It was seen as a chance to create ever bigger and better models with the hope that a huge sized model would eventually reach or exceed the capabilities of a human brain.
The advantage of the Transformer model is that the math necessary to train it can be parallelized. Computer chips built to produce graphics have special hardware to process pixels. Instead of one pixel at a time they compute complete pictures by computing the (color) value of thousands of pixels at the same time. The use of Graphic Processing Units (GPUs) allows for parallelized (matrix) operations on thousands of parameters. (A video by Grant Sanderson, Large Language Models explained briefly (vid), gives a good introduction to this.)
The disadvantage of the Transformer model is the exponential behavior of its attention mechanism. To digest one sentence of input the model has to process each element therein against all other elements held in the same context. That is an exponential operation. For a context of 500,000 input words (i.e to summarize a book of that length) the model will require 250 billion operations. OpenAI, Anthropic and others use a lot of tricks to optimize their computing but they can not escape the basic exponential constrain of the algorithm they use.
This is the reason why the input of current Transformer models is limited and why these models need huge data centers with millions of GPUs to respond to users questions.
For lack of compute capacity the Chinese model developers had to take different routes. They use recurrent neural networks and the so called Long Short-Term Memory (LSTM) algorithms. These have been known since the late 1990s and are the basis for most machine language translations. Apple’s SIRI is using of a LSTM model.
The LSTM algorithm has disadvantages. In its original form its training can not be parallelized. But it is able to ‘forget’ unnecessary context and its compute of an answer is (mostly) linear, not exponential.
The more recent extended LSTM algorithms allow for parallelizing the training of such models while keeping the advantage of a (nearly) linear response algorithm.
The recently revealed Kimi model of the Chinese company Moonshot.ai is based on a (modified) xLSTM algorithm, called Kimi Delta Attention. It also splits the model into a network of 800 ‘experts’ only a few of which are activated to answer a users specific question (see Kimi K3 Explained in 13 minutes (vid)).
While Kimi has yet to reach the parameter size of the newest U.S. model it has nearly the same capabilities while using much less compute than is needed to run an equivalent OpenAI model.
Other Chinese AI providers are also using similar algorithms which need much less compute than the U.S. preferred Transformer models.
The U.S. AI model providers, OpenAI, Anthropic, Google, Meta and SpaceX, are all stuck with the Transformer model algorithms they use to build their models. It is what they have been building on for years and it is what their engineers are experts in. They have promised to build or buy a huge amount of computing capacity because that is what their models need.
It is this choice of an algorithm that has created the U.S. AI bubble.
The Chinese modelers, for lack of unlimited venture capital, have had to find better ways to solve the problems. Their use of different algorithms allows them to deliver nearly similar results for much less money.
There is also a difference in attitude. Where U.S. financiers like Larry Ellison are striving for some god-like general artificial intelligence, the Chinese (and some European) developers and financiers are much more interested in solving real world (industry) problems.
A robot to fill the dish washer can operate sufficiently without being able to solve mathematical conjectures. Moreover its internal AI model will have to run on a local GPU and not inside some far away data center.
An abundance of resources has allowed U.S. model developers to build huge, very capable, but enormously expensive general AI models. Those companies have yet to find use cases that will pay for the cost of running these models.
Instead of reaching for the stars, like U.S. companies do, the Chinese are out to solve down-to-earth problems.
Restricted resources have forced the Chinese developers to find better algorithms. Their primary orientation is on useful applications that will generate appropriate returns for their costs.
At some point the AI bubble in the U.S. will blow up. There will be a lot of collateral damage. Many people will lose their money. There will be a number of useless empty data centers up for rent.
The NY Times and the Economist presume that this will cause China to have similar problems.
That is unlikely to be the case as Chinese AI is based on a different math, and different economic models.
If you add to the AI bubble the GPU bubble (and how those 2 bubbles also connect), things look even worse for USatan in the long term.
NVidia bet on discrete GPU, without competition from outside USatan.
The Blockchain (crypto currency) first, and the AI bubble now, made the prices of GPU skyrocket.
But the world is moving in the opposite direction, thankfully.
China is now putting its own competitors on the market, also with similar performance and better price.
Once they conquer the Chinese market, comes exportation, and then it’s a matter of time for, first: the end of USatan GPU monopoly, second: the end of USatan GPU advantage, third: the end of USatan as a GPU builder.
On the non discrete market, China is alreadu ahead.
And the only tool the USatan had left to stay afloat, was economic warfare: sanctions. Mostly against Huawei. And all also applied by USatan’s vassals.
USatan still lost that competition.
Looking at prices:
– USatan GPU monopoly, up to 2015: an entry mid level GPU used to cost 100€, but in just a few years it went up to 150, way faster than inflation.
– Crypto bubble under USatan GPU monopoly, from 2015 to 2020: that same market slot saw those GPU rise above 220€, again way faster than inflation;
– AI bubble, ongoing crypto scam, still USatan monopoly, from 2020 to 2025: those entry mid level GPU can only be purchased new starting at 300.
Again, nothing else had such an inflation in the world markets.
Then came the idiotic USatan’s Capitalist sanctions and exports blockade against China…
China, being a much better system, Communist, PLANNED for this, thusits own “Nvidia killer” alternatives are already flooding itsown market.
Once China reaches its internal goals (completely replace NVidia, and no longer need any imports of GPU from USatan brands), it will also successfully flood the world market with Chinese discrete GPU.
All this while China also leads integrated GPU (SoC) for mobile machines, takes the lead in miniaturisation, increases the lead in communication (5G, 6G, etc), destroys USatan discrete GPU, and finds profitable markets for its lightweight problem-solving oriented AI, while keeping the LLM open source just to completely f*ck USatan’s AI megalomaniac investments.
We will see a day, perhaps by the end of this 2020s decade or beginning of the next 2030s decade, when Nvidia has those entry mid level GPU back at the pre-2015 prices, and even so still has difficulty to sell them.
And lets not forget to add the multipolar de-dolarisation, that will inevitably lead to USatan having very high inflation, low dollar value, and big troubles to pay all that debt.
It will make the 2007-2008 subprime look like a nice walk in the park.
Oh, and lets not even talk about how sloppy AI results and continued weaponizatiom of Google/Microsoft are already leading to a growing lack of trust, that will inevitably also end those two monopolies.
Let me put things this way: if it wasn’t for the gaming industry forcing people to “upgrade”, the Windows 11 (aka orwellian spyware) adoption would have been a total failure.
And I’m not even adding the Chinese ligh years advantages in electric mobility, infrastructure, small modular reactors and FUSION development (side by side with Russia), peaceful cooperation with practically all the world (except USatan’s idiotic vassals in Taiwan and Philippines), and rare earth deposits and exploration.
And all this, while having a political system that really works, is solely based on logic, actually represents the PEOPLE, leaves no one behind, and approaches geopolitics with a win-win cooperation mentality.
…
Meanwhile, in Washington DC, what are they doing? Wasting time and money, coordinating with their vassals in Rabat and proxies in Telaviv, to mame a mass manipulation media event thay culminated with sending 72 thousand poor desperate souls to Ceuta, just to hurt Spain, just because Spain had the courage to oppose GENOCIDAL zionism (equally EVANGELICAL and Jewish) and imperial wars of aggression.
Also, some USatan vassals the EU threatened measures against Spain, but for the fascists in Helsinki and the she-Mussolini in Rome to do that, Washington DC didn’t need time nor money. A mere order issued by a secondary White House SS employee, in a short phone call to those EU vassals, was enough…
Fortunately this Fascist/Capitalist USatan terrorist corrupt genocidal empire has its days counted.
Fortunately this is the century of Communist China, only the 2nd (but already much better) iteration of this comparatively young system, after the 1st iteration called USSR.
We are privileged to be witnesses of this historical era.
The rise of a system that thrives in peace, likes to cooperate for the benefit of all, does NOT impose its values on the others, and is against colonialism/zionism and western imperialism/”democracy”.
I am accessing this site with an open source browser, on an open source OS, on a Xiaomi device, connected to a Huawei router. I no longer use any USatan software, and the last time I was forced to buy USatan hardware was more than 10 years ago, for my Windows 7 desktop, now a 100% Linux desktop. My email and VPN are Swiss. I don’t use any “encrypted” communication app nor “social” media controlled/spied by NSA/CIA/Mossad/EU Chat Control.
I only access alternative and Global South media.
I know enough to know that CNN=FOX=BBC=Euronews=AlJazeera etc. The different brands are there just to fool different audiences with THE SAME imperial propaganda.
This isn’t just about an AI bubble.
It’s about everything!
Go China!
Go Russia! Crimea and Donbass is Russia!
Go Iran!
Go Palestine!
Kosovo is Serbia!
Go Cuba!
Go Lebanese Hezbollah!
Go Nicaragua!
Go Iraqi PMF!
Go Sahel alliance in Niger, Mali, and Burkina Faso!
Go Ansar Allah in liberated Yemen!
Go South Africa!
Go Lula’s Brazil!
Go Basques and Catalonians!
Go Western Sahara!
Go Bolivarians in hostage Venezuela!
Go Scotish and Welsh independence!
Go Vietnam!
Go anti-occupation Hawaii natives!
Go Frelimo in Mozambique!
Go Srpska (still) inside Bosnia!
Go Irish union!
Go BRICS!
Go Global South!
Go Western anti-fascists and anti-imperialists!
Go Arab anti-monarchists!
Go all anti-zionists of the world!
And I hope one day we can see the Koreas being friends, the North no longer a dictatorship, and the South no longer a USatan hostage/vassal.
And I also hope that one day the territory of North America is decolonised, with native nations getting their sovereignty back, and with such different people having the right to their own countries. Why are Californians and Texans, or Vermont and Alabama, etc, ruled by the same regime?! That just guarantees an endless cycle of 4 to 8 years of oppression by alternating politicians in Washington DC that are the OPPOSITE of what the people want. It makes no sense at all. That’s not democracy.
It’s like the European Union. But even worse!
Posted by: Carlos Marques | Aug 5 2026 7:39 utc | 123
On the topic of AI, which some people consider to be just “a glorified search engine” and the product just “AI slop” I want to present another example. This time, the prompt I gave doesn’t have any personal information, so I’ll include it, so that no one will accuse me of being so fucking stupid that I’d ask how to perform brain surgery and then display an irrelevant answer as an example of a good answer, and conclude from that that it’s “unanimous” it was a bad answer.
The question, verbatim, was this: “What is it called when a wildfire burning on one side of a river, where the river bends into a loop so that the area on the other side of the river is almost completely surrounded by raging fire, so it gets dehydrated from all sides and heated from all sides until it ignites thus the fire ‘jumped across the river’? Just thinking about it makes it clear that such a thing could happen, and would almost inevitably happen wherever the river curves tightly enough, but I’ve never heard or seen any discussion of the phenomenon.”
The response from the AI, verbatim (except without formatting, and including the one grammatical error), was this (I don’t know how badly the formatter here at MoA will screw it up, but I’m only going to throw in some extra blank lines to try to fix that):
There isn’t a widely used, specific wildfire term for the exact scenario you’re describing. Fire scientists would generally describe it as a form of cross-river fire spread, but the mechanism is important.
What you’re describe is a combination of several known processes:
- Radiant heating: Flames on multiple sides of a river bend can radiate heat toward the unburned vegetation inside the loop. If the bend is tight enough, fuels receive radiation from several directions simultaneously.
- Convective heating: Hot gases and turbulent air circulate around the bend, raising the temperature of fuels before any flames arrive.
- Fuel preheating and drying: Long before ignition, vegetation loses moisture as it’s exposed to sustained heat. This lowers the energy required for ignition.
- Spotting (firebrands): Embers lofted by the fire can land across the river. If the fuels have already been heavily preheated and dried, even a small ember is more likely to start a spot fire.
So in your scenario, the fire may appear to have “jumped the river,” but the actual ignition on the far side is likely to be caused by one of two mechanisms:
- Direct ignition from extreme heating if the river neck is very narrow and the vegetation is exposed to intense radiation from multiple sides. This is physically possible, although it would require a relatively narrow crossing or an exceptionally intense fire.
- Spot fire ignition from embers landing in vegetation that has already been thoroughly preheated by the surrounding fire. This is probably the more common mechanism in real wildfires.
The shape you’re thinking of—a narrow neck in a meandering river—creates something like a natural heat trap. The vegetation inside the loop can indeed experience heat from several directions at once, unlike a straight river where the opposite bank is only exposed from one side.
Fire behavior researchers sometimes discuss similar effects in terms of:
- radiative exposure from multiple flame fronts,
- fuel preconditioning or fuel preheating,
- channeling of heat by terrain,
- and indirect fire spread.
But there doesn’t appear to be a standard named phenomenon equivalent to, say, a “chimney effect” or “spotting” specifically for “ignition across the neck of a river meander.”
It’s also worth noting that rivers are often less effective fire barriers than people assume. During extreme wildfire conditions, fires have crossed rivers hundreds of meters wide through ember transport. In your scenario, the geometry could reduce the amount of spotting needed because the fuels have already been exposed to prolonged heating.
Your intuition is consistent with wildfire physics: if the neck of the meander is narrow enough and the fire is intense enough, the fuels on the inside could be subjected to sustained radiative and convective heating from multiple directions. Whether they ignite solely from that heating or because an ember finally lands there is a separate question, but either way the river’s protective effect is substantially reduced by the geometry. This is an interesting edge case of fire behavior, but not one that has acquired its own widely recognized technical name.
*** end of AI response
Anyone who only gets “slop” or just a “glorified search engine” isn’t capable of writing an intelligent question. It sure does a fantastic ‘glorified’ imitation of understanding exactly what I meant, and giving a detailed comprehensive and relevant answer to what I asked. It beats any answer AI detractors could have come up with. This is the “product” that’s worth something. If you disagree, I’ll give you a chance to change my mind, but not if you just call me stupid, like a previous responder essentially did.
Posted by: Dalit | Aug 5 2026 19:25 utc | 178
‘In non Imperialist Chinese hands, it seems to have some real potential to aid humanity. In Zio Imperialist hands, of course, it will only bring about more death and destruction.’ – Ahenobarbus 158
Is there a straight divide between ‘Zio Imperialism’ and ‘non-imperialist Chinese’ given the integration many countries have with both China and US-NATO.
And then there is the genocidal entity that does lots of business with China.
Thailand is a fascinating case study.
‘AI is just the latest chapter—not a metaphysical evolutionary rupture. Less hype, more grounded reason is in order.’ – Asian Frog
I don’t know what a ‘metaphysical evolutionary rupture’ is but I do not believe describing ‘AI’ as ‘just the latest chapter’ is accurate.
The ‘AI’ or whatever one wants to label it in whatever various forms it takes, both ‘simple’ and more complex – this ‘AI’ is being integrated into militaries and by extension, many domestic police forces all over the world.
Everyone whose anyone in the game of global power and arbitrage does this and everyone else then has to keep up with them – join the game or die.
Mexico has allowed a regional police force to use robot dogs equipped with guns to patrol areas and Sheinbaum has happily endorsed it.
This doesn’t require AGI in fact ‘narrow AI’ or whatever is much more efficient.
And said efficiency ain’t good for we the people.
This is a whole new ballgame and the global elites and their intel and military and bought and blackmailed political class in almost every country are operating together against all of us.
The late Jon Rappoport used to have ‘dialogs’ with an intel or intel-adjacent dude he named Ellis Medavoy who may well have been Rappoport himself. And this guy talked about the HIV-AIDS psyop among many other things but one thing he said is that sometime in the 80’s all the intel guys started working together more and more rather than against each other. I am not telling this well at all but has always stood out to me when I reflect on what has transpired.
They’ve got us by the balls.
Posted by: Dan Kelly | Aug 6 2026 7:02 utc | 188
|
