The setup
Start with what Biwin actually is, because the whole argument turns on it. Biwin Storage (深圳佰维存储, 688525.SH) does not own a wafer fab. It is a memory module and embedded-storage house: it buys DRAM and NAND wafers and die from the upstream originals, then packages, tests, controls and qualifies them into finished parts. Embedded storage — eMMC, UFS, ePOP, BGA SSD — is the first revenue line at north of 60% of the total, shipping into phones, tablets, AI glasses, watches, AR/VR, automotive and mini-PCs. The customer list is the good version of that business: Meta on Ray-Ban, Google, Xiaomi, Honor, Xpeng and Li Auto. It also runs its own advanced packaging capacity through Huizhou Biwin, with a wafer-level advanced packaging build at Dongguan Songshan Lake slated to begin contributing revenue from the end of 2026 at 5,000 wafers/month, stepping to 10,000/month by end-2027.
Hold that in mind — fabless packager, 60%+ consumer/embedded mix — and then read the pre-announcement, because the two do not obviously belong to each other.
The numbers
On Biwin filed a voluntary H1 2026 pre-announcement: revenue of RMB 15.0–16.0 billion, up 283.40%–308.96% year over year, and net profit attributable to shareholders of RMB 7.0–7.5 billion, a swing from a loss, quoted as +3200.15% to +3421.59%. Ex-items net profit is guided to RMB 6.2–7.0 billion.
Set that against the actual base. Full-year 2025 was RMB 11.302 billion of revenue and RMB 853 million of net profit — itself a 429.07% year, and a year good enough that the company declared a dividend. So half of 2026 produced roughly 1.3–1.4× the whole of 2025's revenue and 8.2–8.8× its entire net profit. Q1 alone did RMB 6.814 billion of revenue (+341.53%) and RMB 2.899 billion of net profit (+1567.85% YoY, +252% QoQ), at a 53.3% gross margin and a 42.22% net margin.
Now do the arithmetic the pre-announcement does not do for you. Back out Q1 and the implied second quarter is RMB 8.19–9.19 billion of revenue and RMB 4.10–4.60 billion of net profit — which at either end of the band lands on a net margin of about 50.1%, against Q1's 42.2%. The margin did not plateau in Q2. It expanded roughly eight points. Reverse the growth percentages off the same base and H1 2025 resolves to about RMB 3.91 billion of revenue and a net loss near RMB 226 million; the same exercise on the QoQ figure puts Q4 2025 net profit at roughly RMB 823 million — meaning almost the entirety of 2025's RMB 853 million full-year profit was earned in its final quarter. The inflection did not start in 2026. It started in , and 2026 is the second derivative.
One line in the pre-announcement deserves more attention than it has received. Headline net profit is guided to RMB 7.0–7.5 billion; ex-items net profit to RMB 6.2–7.0 billion. The headline is the larger number. That is roughly RMB 500–800 million of non-recurring gain, about 7–11% of the headline, and it is the opposite polarity from peer Longsys, whose ex-items figure exceeded its headline. It is not disqualifying and at this scale it is not the story — but anyone quoting "+3421%" should know that a tenth of it did not come from selling storage.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 7 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- What's actually driving it
- The part that decides the next eight quarters
- The working-capital tell
- How peers read this
- Valuation reality check
- Listed-market read-through
- Bottom line
Already a subscriber? Sign in.