Tenant Dapuw

I recently came across a widely circulated claim.

"Dapu Micro, a company that only makes hard drives, why is it worth so much money?" — at first glance, this statement sounds reasonable. An SSD, a solid-state drive, is after all just a storage device. Why would it be so expensive?

01

Hold on. To answer "why," you first need to understand the position this company occupies.

Let me put it this way to make it clear:

It's like a tenant whose very livelihood depends on a piece of land that it doesn't own — that land is leased from its arch-rival. Then, over the course of ten years, it used that rented land to poach customers from its rival, a fact that's clearly documented in its financial reports.

Dapu Micro focuses almost entirely on one thing: enterprise-grade SSDs for data centers.

In 2026, the company stated on an interactive platform that enterprise-grade SSDs account for 98.98% of its main revenue. All the remaining revenue from technical services, RAID cards, and other miscellaneous items combined doesn't even make up a tiny fraction of that total.

Note that I'm not talking about consumer-grade products. The Samsung 980 Pro SSD you plug into your computer follows a completely different product logic from the drives that Dapu Micro sells to ByteDance and Google.

Consumer-grade SSDs are judged by benchmark scores and price, while enterprise-grade SSDs are judged by stability and consistency — whether they can run 24/7 on hundreds of thousands of servers for three to five years without any failures.

How narrow is this threshold?

The validation cycle for an enterprise customer is 6 to 18 months. From sending samples to bulk procurement, there are endless rounds of testing covering compatibility, reliability, power consumption, and firmware stability, with each round taking weeks to months.

Most suppliers get stuck right outside this gate. Once you squeeze through, competitors who want to get in will have to start all over again, facing another 6 to 18 months of checks.

How did Dapu Micro manage to get in?

From my research, it develops its own controller chips — the PCIe 4.0 DP600 and the PCIe 5.0 DP800. It also develops its own firmware algorithms, and has the capabilities for module design and verification. In the industry, this is called full-stack in-house development.

In plain terms, when a customer proposes a scenario-specific requirement, the company doesn't have to wait for upstream suppliers to modify chips or external parties to adjust firmware — it can make all the tweaks on its own.

The numbers don't lie — it truly broke through that barrier.

In 2025, overseas revenue reached 1.065 billion yuan, accounting for 46.55% of total revenue, with a year-on-year increase of 656%.

It has already started bulk shipments to Google. In 2025, it also passed the testing and onboarding processes of two global AI leaders, Nvidia and xAI. While it's hard to say how big the business will eventually grow, the door is already open. In China, you can count the companies that have obtained this admission ticket on one hand.

At this point, the first half of the answer to "why it's valuable" becomes clear: it has successfully secured those customers.

But what about the land it leases?

On July 8, 2026, Dapu Micro confirmed on an interactive platform that original manufacturers including SK Hynix are its suppliers of NAND Flash and DRAM, and it purchases components through authorized distributors.

Here's the problem: one of Dapu Micro's most direct competitors in the international market is Solidigm, which is a subsidiary of SK Hynix.

This is exactly that piece of rented land. On one hand, the company buys memory chips from its competitor's parent company, and on the other hand, it fights for market share with that same competitor in front of clients. The "house" is fully built by itself — its in-house developed controller and firmware are genuine, solid technology. But the "land" is rented from its rival.

The cost is straightforward: it has no control over the pricing of upstream wafers. There's a trap in its procurement contracts: purchase volumes are locked in, but prices fluctuate with the market. When memory chip prices rise, its costs go up. When prices fall, the value of its inventory drops.

Simply put, it has to accept rent hikes, and it has to bear losses if the value of its "house" depreciates.

The company went public on the ChiNext board on April 16, 2026, becoming the first unprofitable enterprise to list on that board.

Its stock ticker has two letters at the end: W and U. W stands for weighted voting rights, meaning the founding team has greater voting power to maintain control of the company. U simply means the company is still unprofitable.

Let's look at a set of figures:

From 2022 to 2024, its revenue was 557 million yuan, 519 million yuan, and 962 million yuan respectively. What about its net profit attributable to shareholders? -534 million yuan, -617 million yuan, and -191 million yuan. Over three years, while revenue climbed from 500 million yuan to nearly 1 billion yuan, the company's accumulated losses grew deeper. By the end of 2025, its retained earnings were -1.071 billion yuan.

So what kind of company is this? It leases land from its rival, has already secured a group of customers, holds firm control of the company, and hasn't started making profits yet.

Once you understand this identity, the question naturally arises: why would the market assign such a high valuation to an unprofitable company?

02

After doing my research, I realize the market isn't that irrational — it's calculating a different kind of value.

Let's start with a few sets of data:

For the full year of 2025, revenue was 2.289 billion yuan with a net loss of 481 million yuan. But by the first quarter of 2026, quarterly revenue hit 1.31 billion yuan, a year-on-year increase of 340.95%. Its net profit attributable to shareholders was 370 million yuan, marking a year-on-year return to profitability.

Notice that in the same period the previous year, the company still posted a loss of 124 million yuan. The profit it earned in one quarter was enough to reverse all its losses from previous years combined. This is called an inflection point, and that inflection point has a name: the AI storage price surge cycle.

In the second quarter of 2026, the contract price of NAND Flash rose by about 70% quarter-on-quarter, and DRAM prices rose by around 60%. The entire storage industry is experiencing a price surge that ranks among the most significant in a decade.

Dapu Micro itself stated in its prospectus:

It expects the unit price of its main products to rise by 30% to 40% in 2026, driving full-year revenue to surge sharply to 3.357 billion yuan. If prices rise by 30-40% while costs grow at a slower pace, gross margins will naturally expand.

What was Dapu Micro's gross margin in the first quarter? It reached 37.56%.

Let me make a comparison for you: Module manufacturers typically have single-digit gross margins. For every 100 yuan of goods sold, a module manufacturer earns 5 yuan, but Dapu Micro earns 37 yuan — more than seven times the difference.

Earlier, we described it as a tenant building a house on rented land to explain its situation, but now we're talking about how it generates profits.

You can think of it as an amplifier: its input is in-house developed controllers and contracts with top-tier clients, and its output is gross margins and profits. The stronger the input, the greater the amplification of the output.

Module manufacturers profit from inventory price differences, buying memory chips at low prices and selling finished products at high prices. But Dapu Micro profits from technology premium: its in-house developed controller and firmware allow its products to command higher prices from leading clients than generic alternatives.

One is a wholesaler earning from price differences, the other is a branded manufacturer profiting from moat-based premiums.

When a price surge cycle arrives, this amplifier magnifies the technology premium. Even with the same 30% price increase, a module manufacturer might see its single-digit gross margin rise by a few percentage points, while Dapu Micro's margin could jump from 37% toward 50%.

Who benefits more? Dapu Micro does. This is the upward amplification effect of the "machine."

Add to that the long-term contracts it secured earlier — bulk shipments to Google and onboarding with Nvidia. Once a supplier is integrated, customers won't switch to a new one just because of a price increase.

Switching suppliers means restarting that 18-month validation process, which no one can afford. During a price surge cycle, these contracts themselves become additional amplifiers.

So what's the market valuation?

By the close of trading on July 6, 2026, its total market capitalization was about 261.5 billion yuan, with a dynamic P/E ratio of 176.76x and a trailing P/E ratio of 20,000x.

According to the China Securities Index Co., Ltd., the industry it belongs to — "computer, communications, and other electronic equipment manufacturing" — has an average trailing P/E ratio of around 58x. The premium the market has assigned to this "amplifier" is entirely reflected in that massive gap.

A 20,000x trailing P/E ratio sounds crazy, but the math is actually misleading. Most of the four quarters used to calculate that figure were still loss-making, so dividing by a tiny profit base naturally gives an absurd number. The real bet is on the dynamic P/E ratio.

Annualizing its first-quarter net profit of 370 million yuan gives a dynamic P/E ratio of roughly 176x relative to its market capitalization.

Among consensus estimates from sell-side analysts, only one brokerage released a coverage report in May 2026, forecasting full-year net profit of about 2.726 billion yuan. If that number holds, the dynamic P/E ratio would drop to around 96x.

So the market is betting on "how much this amplifier can multiply profits during the price surge cycle." But this calculation has a prerequisite: the cycle must keep going up. What if the price rise stops, or reverses direction?

03

The company itself stated very directly in a risk warning announcement:

AI-driven demand for enterprise-grade SSDs is subject to volatility and uncertainty due to the pace of AI commercialization, capital expenditures of cloud vendors, and downstream industry prosperity. In other words: the price surge might not last as long as expected, and the downturn could come faster than you think.

What does this amplifier look like when the tide recedes?

Look at the fourth quarter of 2025: its quarterly gross margin was negative. For every 100 yuan of goods sold, it not only made no profit, but actually lost money to get the products out the door. Its peers' gross margins also fell, but none of them dropped into negative territory. Why did Dapu Micro's margin turn negative?

Two factors overlapped.

First, it didn't hold enough inventory. By the end of 2025, Dapu Micro's inventory was about 1.45 billion yuan. In the same period, Jiangbolong had 17.961 billion yuan, Demingli had 12.192 billion yuan, and Baiwei Storage had 12.069 billion yuan — nearly ten times more inventory. They're not even in the same league.

Second, the inventory it did have was purchased at a premium price.

The company itself explained in a reply to a regulatory inquiry letter that during the storage market volatility in the fourth quarter of 2024, its weighted average procurement cost for NAND Flash was relatively high. When it sold this high-cost inventory in 2025, the cost of goods naturally ran high.

Low inventory combined with high-cost inventory created a double squeeze, pushing gross margin into negative territory. This is the downward amplification effect of the "machine."

During the price surge cycle, module manufacturers reaped huge profits while Dapu Micro had to take a hit before getting a share of the gains. It has huge upside elasticity, but when the market falls, it crashes harder. Since it doesn't control its supply base, the direction of the amplifier's gain is not up to the company.

There's another set of figures that paints a clearer picture than just the 3.0% market share number.

According to IDC's "2024 China Enterprise Solid-State Drive Market Tracking Report," the total market size in 2024 reached 6.25 billion US dollars, surging 187.9% year-on-year — almost tripling in size.

What about the rankings? Solidigm and Samsung continued to lead the market.

Yilian held an 11.4% market share, remaining the top domestic player and third in the industry, with shipments up 117% year-on-year. Yiheng Chuangyuan ranked fourth, and Dapu Micro came in fifth with 3.0%.

The market nearly tripled in size, but its market share dropped from 6.4% to 3.0%. It lost market share not just to overseas giants, but also to domestic competitors Yilian and Yiheng Chuangyuan.

It failed to keep pace with its peers during the boom, so when the tide recedes, there's a big question mark over how much its moat can actually protect.

Overseas growth has been rapid, but it comes at a cost: in 2025, its overseas gross margin was only 5.50%, while its domestic gross margin was 1.78%. Its overseas business has large volume but thin margins — it's growing in scale, not in profitability.

Putting these five points together: low inventory, locked volumes but flexible pricing, a history of negative gross margins, declining market share, and low-margin overseas business. What's the company's own judgment? It expects storage prices to keep rising throughout 2026, and to achieve overall profitability by the earliest possible time that year.

From my research, only one brokerage in the entire market has provided a full-year profit forecast of just over 2.7 billion yuan. Just one. Mainstream institutional investors are still on the sidelines.

If the price cycle can't hold up, this amplifier will reverse to magnify losses. That's the inherent design of this "machine," and also the inherent design of that rented land.

04

We've talked about the machine, now let's talk about the people. Who's in the game? Who's getting out? Who's waiting on the sidelines? Three groups of people, laid out clearly — you can make your own judgment.

The founder, Yang Yafei, born in 1979, holds a PhD. He worked at Qualcomm in the United States for eight years, rising from senior engineer to senior staff engineer. He returned to China in 2016 to found Dapu Micro. Through Dapu Haide and Dapu Haiju, he indirectly holds a total of 11.15% of the shares.

A 11.15% stake isn't high in the A-share market, but notice that W at the end of the stock ticker — the weighted voting rights structure means that each special voting share has 10 times the voting power of an ordinary share.