Cheap AI for most, frontier AI for special use

Over the past week, I have been exploring the changing landscape of AI the Chinese brought out free open-source AI.

The paradigm suddenly shifted from the AI hyperscalers monopolizing AI with expensive, full-capability “frontier” AI to a business model where small, customized, “distilled” AI programs can handle >90% of routine applications cheaply and often in-house.

https://www.wsj.com/business/china-us-ai-model-costs-53a12e96?mod=hp_lead_pos7

Corporate America Has Suddenly Decided to Stop Blowing Money on AI

Companies big and small are mixing models and it’s changing the economics and power players of the industry

By Angel Au-Yeung,Katherine Bindley and Tina Li, The Wall Street Journal, July 24, 2026

Companies across the country are coming around to a radical idea with the potential to upend the industry powering the global economy: They don’t have to blow their budgets on AI.

Fed up with ballooning costs, companies big and small are starting to use lower-priced models, including some built in China. In many cases, they are adding the new, cheaper models alongside OpenAI and Anthropic’s products, shopping a la carte for their artificial intelligence…

The most powerful and expensive AI models aren’t necessary for relatively mundane tasks.

“It’s like driving a Lamborghini to go to the grocery store to pick up milk when that was designed to be raced around a track…”

Being economical—or tokenomical—is a dramatic reversal in mindset. Just a few months ago, it was a badge of honor to be using AI so much that you spent a lot on tokens. Companies rewarded employees for tokenmaxxing, flashing leaderboards that showed who had spent the most. Now they are thrift-maxxing.

The shift in their budgeting isn’t just about how much U.S. companies are spending on AI. It’s also a geopolitical issue that pits the world’s economic superpowers against each other. Generally, the best-known U.S. models are closed, which means they are strictly controlled by the companies developing them. China is known for cheaper and open-weight models, which means they can be downloaded and customized…

Strategies to lower AI costs include limiting access to top models for new hires and using the most advanced AI systems to plan how tasks will be completed before turning to cheaper models for the execution… [end quote]

This new paradigm is already causing the hyperscalers to try to lock in customers, offering partnerships, tens of thousands of dollars in incentives and heavily subsidized AI usage. That will cut into the high earnings growth expectations that justify the high P/E multiples.

The new paradigm also cuts into the expected use forecast since much of the AI “inference” will be done by distilled models instead of frontier models. Giant data centers that were built to satisfy the demand for frontier AI could have excess capacity, driving down price. (Typical of the pricing cycles of semiconductor fabs.)

Responsible CTOs and CFOs will immediately jump on this opportunity to save their company money while improving productivity with AI that’s designed for their specific needs.

This thrifty paradigm disrupts the business model that supported the AI bubble by bridging the wide moat the hyperscalers (and their investors) counted on…only a week ago.
Wendy

2 Likes

And once again our focus on short term profits causes us to let China dominate the next emerging market. Really clever.

Reminds us of what happened w consumer electronics where companies like RCA once dominated.

4 Likes

It is not like the Chinese AI companies are making money.

Be my guest.

They don’t have to make money to achieve a Chinese government objective: messing up the American stock market.

Wendy

3 Likes

China might make money.

See Freiberg’s comments beginning about 41 minutes.

Essentially, Freiberg says commoditizing the software reduces the “value” (the profits) gained from that software.

China, by open sourcing cheap models is depriving US (n West) of that wealth.

Meanwhile China continues as the “world’s factory”, converting the atoms into usable hardware. And selling that hardware at prices that US n West companies cannot match.

Ie, China profits from commoditizing the AI software. Both in terms of economic gain and in terms of influence (hegemony).

Freiberg says that China, as the world’s factory, requires energy and atoms.
The energy allows the atoms to be processed into physical hardware.

What is the goal of taking over VZ n Iran?
VZ n Iran were major oil (energy) exporters, that energy was sanctioned and China was buying black market energy at below market price.

Shutting down the flow of cheap oil to China affects China’s ability to manufacture stuff at prices below Western companies.

China is affecting Western AI economics; the West (US) is affecting China’s manufacturing economics.

This, IMO, is at a higher level of international/global macroeconomics than the “regional economics” of VZ (Cuba, Monroe Doctrine); and Iran (IRGC nukes).
But is the driving force behind those current conflicts.

:speak_no_evil_monkey::hear_no_evil_monkey::see_no_evil_monkey:
ralph

Commoditizing “something” in order to shutdown a competitor’s income and maintain market share, is time honored.

US reshoring of manufacturing, and EU attempts to appease the Chinese into “playing nice”, are attempts to level the “manufacturing” playing field.

2 Likes

The Chinese companies making solar panels weren’t/aren’t making money, but you can barely find solar panels made in this country because all the work went to China.

Chinese companies making EVs weren’t/aren’t making money, but they are now world class manufacturers, and the rest of the countries on the planet are worried about their automotive industries.

You know China makes 70% of the display screens in the world? You know how they did that? By not making money on a strategic industry in the beginning. (Did you know the US military came to rely on Chinese displays for fighter jet cockpits until that was outlawed by legislation.)

China builds more than half the heavy shipping in the world thanks to state subsidies. You know how many the US builds? 0.1%

Want to buy a lithium battery for home storage, EVs, or other uses? China produces 80% of them. Not at a profit.

I have said before: China is thinking of the next 25 years. US industries, with their short term focus are thinking of the next 25 weeks.

9 Likes

That’s not really the case. Last year the US installed 26.5 GW of solar capacity.

At the same time we read that US solar manufacturing capacity is great than 50 GW.

The US imports very few solar panels from China due to high tariffs and trade restrictions.

DB2

4 Likes

If that was true, how come China has deflation? That is the worst of economic failures.

BTW we are shutting off as much oil as possible. We are not far behind in getting the deflation bug.