'When the chips are down' — what a difference 7 years can make!

Seven years ago, I stepped off the train at Hefei South Railway Station in Anhui province to visit a memory chip startup that had just developed its first self-designed high-end product. At the time, the global DRAM industry was a near-monopoly of Samsung, SK Hynix and Micron. Few believed a Chinese upstart could ever crash that party.

On July 27, China's largest memory chipmaker ChangXin Memory Technologies, or CXMT, made a blockbuster debut on the Shanghai Stock Exchange's STAR Market. Shares surged 465.82 percent to close at 49 yuan, propelling its market value to more than 3 trillion yuan ($444.9 billion). The company not only overtook US peer Intel in market value, but also dethroned the Industrial and Commercial Bank of China as the domestic market's top-weighted company. The single-day turnover of 140 billion yuan shattered all records for an individual A-share stock.

The seven-year arc from my first site visit to today's financial milestone is not just a corporate success story. It is a testament to what happens when strategic patience meets a fortuitous market moment.

To understand CXMT's rise, one must first understand its founder. In 2005, Zhu Yiming founded GigaDevice Semiconductor Inc, a Nor Flash memory company, with just $100,000 in startup capital. He took it public on Shanghai's main board in August 2016 — a comfortable exit by any measure. At 44, he was wealthy, respected and could have coasted on GigaDevice's success.

But Zhu had a bigger ambition. Nor Flash was not mainstream memory. China, the world's largest electronics manufacturing base, consumed enormous quantities of DRAM chips each year — yet had no meaningful domestic production capability.

Zhu knew China had to crack DRAM. So in 2016, he made a choice that few understood: he plunged into what would become a decade-long, capital-intensive, and brutally difficult war of attrition.

On May 6, 2016, Zhu reached an agreement with the city of Hefei to launch CXMT. The total commitment would eventually reach 150 billion yuan.

In 2019, CXMT achieved what no Chinese company had done before: it successfully mass-produced China's first self-designed, self-manufactured 8-gigabyte DDR4 memory chip. It was a historic "from zero to one" breakthrough for China's DRAM industry.

That's exactly when I first visited the company. I still remember the pride the engineers had when they showed us their first products, and I even brought home a souvenir version of the packaged 8GB DDR4 memory chip.

It is worth noting, however, that CXMT is not the only Chinese company that has tried to develop DRAM chips. Tsinghua Unigroup, a major Chinese integrated circuit manufacturer, also spent big money on building factories and acquiring talent from the bankrupt German DRAM maker Qimonda, but it failed.

DRAM is a scale-driven, capital-intensive business with brutal technology barriers. CXMT's success is a combination of long-term capital investment, persistent technological innovation and good timing.

In fact, from 2016 through 2025, CXMT burned through cash year after year. According to its prospectus, as of Dec 31, 2025, the company's cumulative losses had reached 36.65 billion yuan.

Then came 2025 — the turning point.

Driven by explosive artificial intelligence demand, the global DRAM market entered a historic super-cycle. Hyperscalers ramped up infrastructure spending, and server demand for memory chips surged. CXMT's capacity expansion — three 12-inch fabs across Hefei and Beijing — hit the market at precisely the right moment.

The numbers tell the story. In 2025, CXMT's revenue reached 61.799 billion yuan, up 155.6 percent year-on-year, with net profit attributable to the parent company of 1.875 billion yuan — its first full-year profit.

Then 2026 blew those numbers away. In the first quarter alone, CXMT generated 50.8 billion yuan in revenue — up 719 percent year-on-year — and net profit attributable to the parent company of 24.762 billion yuan.

By the second quarter of 2026, CXMT's global DRAM market share reached 7 percent, ranking fourth worldwide, according to the latest data from global technology market intelligence company Counterpoint Research. It has broken the three-way monopoly that once seemed unassailable.

When I asked a veteran engineer what the IPO means for the team, he smiled: "Seven years ago, we were proving we could make a chip. Now we're proving we can build a business."

Some will say CXMT is simply riding a wave. True, global DRAM revenue more than doubled over the past year, but windfalls don't explain everything. CXMT didn't just get lucky; it positioned itself to capitalize. Its long-term research and development strategy paid off. Its capacity expansion hit the market at exactly the right moment.

During the IPO roadshow, Zhu said the company would "actively seize the incremental opportunities brought by AI computing power" while remaining "objective about the strong cyclical nature of the DRAM industry." That is the voice of someone who has learned to dance with volatility.

The question is no longer whether China can compete. It is how fast it can catch up — and what happens when it does.