The Eve of the AI Bubble Bursting

Today is Saturday, October 31, 2026. The US market just closed on Friday.

The Nasdaq fell 2.86%. Not a terrifying number by any means, but the internet is full of wailing and gnashing of teeth.

Intel dropped 28.58% on Friday — its biggest single-day decline in forty years. Its Q3 EPS came in at 0.52, far exceeding the market consensus of 0.40.

Just yesterday, Google fell 12.64%, closing at $283.82 — its worst day since its 2004 IPO. Nearly $500 billion in market cap evaporated in a single day. But its revenue missed expectations by only 0.2% to 0.4%. Its cloud revenue doubled, and search query volume growth hit a five-year high.

SanDisk’s revenue grew 23% quarter-over-quarter. Adjusted EPS of $49.51 — up more than 40x year-over-year. It has $91.1 billion in contracted orders on the books. Earnings beat expectations, and the stock still fell 13.47%.

When all of this happens, the market’s first reaction is: the AI bubble is about to burst.

There certainly is an AI bubble, and it will certainly burst. Whether it’s bursting today, I don’t know — and it doesn’t matter.

This article is not investment advice, and investing is not the topic I want to discuss today.

I want to talk about whether AI is actually creating incremental value right now.

AI’s Value Is Hard to Make Sense Of

The thing that’s had the biggest impact on us this year is probably the price surge in memory and flash storage.

Hardware price increases have directly led to declines in PC and smartphone sales. Memory now accounts for 40% of the cost of an iPhone. Apple didn’t even release a standard iPhone this generation.

The new iPad mini sells for ¥5,499, but only comes with 128GB of storage — not even enough for two large games.

Many Android manufacturers have also stopped releasing budget phones under ¥1,000.

Intel’s forecast: the PC market will shrink 12% in 2026 and another 10% in 2027, with the downturn continuing.

We were told AI would make us happier. Instead, we can no longer afford phones and computers.

The root cause of rising hardware prices is the great leap forward in AI cloud infrastructure.

AI needs cloud hardware for inference, and hardware for human use is dragged up in price.

Sam [Altman] said that running local AI on personal computers is a waste of memory. The implication: those chips should go to AI, creating greater value.

He’s speaking entirely from the perspective of a leading AI model company. The value he sees is only the value he believes in.

Of course AI is powerful and valuable. No one doubts its economic value.

And precisely because AI is so powerful, it can replace countless white-collar jobs. That substitution value is beyond dispute.

But beyond substitution value — what incremental value has AI created?

This is what I find most incredible.

When I think about what incremental value AI has actually brought…

I find that there is none.

At least not at this point in time — October 31, 2026.

If there is one exception, it’s Hongguo short drama (红果短剧). But even that is bizarre. I’ll get to it at the end.

Is AI a “Treat Yourself” Economy?

I know this conclusion will draw a lot of pushback. Let’s look at AI’s strongest domain: coding. AI coding is so powerful — how can you say it hasn’t created incremental value?

Yes, AI’s coding capabilities are beyond question. Every software company has seen coding productivity boosted by AI. Code output at every company is exploding — one year equaling the past ten.

But productivity is not the same as value.

AI coding boosts productivity 10x, but it hasn’t increased product revenue by 10x. Even 2x is extremely rare.

Swap in another example and the point becomes clear.

Yunnan’s blueberry output grew 39% this year compared to last, yet prices dropped 50%.

The official conclusion: “A flood of fresh fruit hit the market at once. A shift in the supply-demand structure caused prices to collapse.”

Increased supply doesn’t necessarily mean increased revenue — because things got cheaper.

Economics is ruthless that way. Scarcity creates value. Humans pay for scarcity.

Prices are determined by supply and demand. When demand stays constant, more supply means lower prices.

Last year, you built a business that earned you $1 million per year.

This year, you have more powerful AI. You think: great, productivity is up — this year I’ll earn $10 million!

But don’t forget: AI lets your competitors replicate your entire business in a day. The number of competitors has therefore increased 10x. The result is fiercer competition, and consumers get cheaper goods and services.

And your life just got harder.

What AI does is make intelligence no longer scarce.

No scarcity means no value.

The tokens that drive intelligence are getting cheaper.

Products built on intelligence are getting cheaper.

Jobs based on intellectual activity are getting cheaper.

To offset this decline, we must increase volume. Consume 10x more tokens, produce 10x more products, work 10x harder. And that only barely maintains the balance of value.

This year, I’ve watched my friends frantically use AI to write code. They optimize every detail to squeeze the most out of resets. Product after product after product ships.

But are they making money? Not a single one.

AI has indeed made tasks that weren’t previously worth paying for suddenly achievable at the tap of a button.

But AI doesn’t seem to have created incremental value.

I asked my friends whether their vibe coding is saving time or killing time. They all said killing time.

No wonder someone said the essence of AI isn’t B2B, and it isn’t B2C — it’s 2ME. Doing it for your own enjoyment.

So it turns out it’s a “treat yourself” economy.

In the AI Economy, There Are No Middlemen

The most bizarre thing about the AI economy: there are no middlemen skimming the spread.

Wait — isn’t that a good thing?

The basic principle of economics is about transactions. Transactions make everyone’s life better.

One person earns money through transactions, then spends it — and someone else earns money, and they spend it. Round and round, endlessly.

But AI is anti-transaction. A person gets results directly from AI that will soon cost as little as electricity. Almost no other human earns money in the process. And so others can’t spend that money.

AI replaces the jobs of the people who do the consuming — so who’s left to consume?

Even the logic of content consumption has changed.

Reddit just had its worst day since going public — the stock fell 29%.

Reddit CEO Steve Huffman’s explanation on the earnings call was simple:

Google’s AI Mode now delivers answers directly. Nobody clicks links anymore.

“AI doesn’t bring traffic. Not to us, and not to anyone else.”

For the past 25 years, the internet operated on this protocol: websites provide content to Google, and Google sends users back to websites.

But now AI Mode summarizes everything, and users get their answer and leave.

In October, average weekly visits from search engines dropped by over 1 million. Daily users entering websites in the US fell 2% year-over-year — the first decline since Reddit’s 2024 IPO.

Reddit is just the tip of the iceberg. It’s widely considered the most “human” community online. If it’s affected, other sites are far worse off.

No egg survives when the nest is overturned.

But aren’t middlemen (API relay/proxy services) still profitable? How can you say there are none?

From Claude Opus 4.5 to DeepSeek Flash 4.1, the token price for completing the same task dropped 50–100x this year — and it’s expected to drop at least another 10x next year.

There are hardly any relay services offering low-priced models like Luna anymore. A friend who runs a relay service told me: revenue looks decent on paper, but after deducting costs, the profit is minimal.

Tokens are too cheap — cheaper than the cost of running the relay servers.

All the Money Humans Earn Is Going to AI

There was another major development this week: Microsoft offered voluntary buyouts to 60,000 employees — more than a quarter of its workforce.

Everyone knows this is a gentle form of layoff.

Is Microsoft in trouble? Why would it want to cut a quarter of its staff?

The reality is this company is making more money than ever — and it’s still pouring that money into AI investment.

Not enough money to invest in AI? Lay off employees to fund it.

This brings us to ByteDance’s full loop: Hongguo short drama.

I have to say, Hongguo is an even more impressive product than Douyin. Short dramas for everyone, young and old.

Fanqie Novel’s text + Seedance’s video generation + Ocean Engine’s intelligent ads + Hongguo’s DAU and engagement duration.

From creator, to model, to recommendation, to consumer — fully closed loop.

That should be incredibly profitable.

But where did ByteDance’s money go?

ByteDance earned approximately $50 billion last year. This year, it plans to spend up to $70 billion building AI infrastructure.

ByteDance is putting most of its cash into AI.

Do you see the pattern? Hardware investment goes to AI. Productivity software companies spend their money on AI. Even with a fully closed consumer loop, the money still goes to AI.

We live in an era of blind, reckless AI investment.

We’ve gone all-in on productivity without any consideration for relations of production.

After the Bubble

I feel like we’re standing on the eve of an AI bubble bursting.

I don’t think today is necessarily Black Friday, and I don’t know when this AI bubble will pop.

Looking at history, in every technology cycle, a technology first goes through a bubble, then suffers a major crash to deflate it — and only then do people start putting it to genuinely valuable use.

That’s the cycle. Behind the cycle is human nature.

Technology always changes. Human nature never does.

From ChatGPT’s launch in November 2022 to today, AI is about to complete a full four years.

Anthropic is going public next month. Many say that will mark the peak of the AI bubble.

What do you think?

Note: This article is based solely on publicly available information. The author knows nothing about investing. This is not investment advice.