Nvidia has made a point of it in three consecutive earnings decks: year-on-year growth has accelerated for four straight quarters, which the company calls unprecedented at its scale. It's true. But split the data center business into its two halves, and the acceleration has a single source, and it isn't the customers everyone assumes.

Yangtze Memory Technologies (YMTC) has officially kicked off its STAR Market IPO process. In the first quarter of 2026, China's largest 3D NAND Flash manufacturer delivered an impressive performance, generating CNY47.042 billion (approx. US$7 billion) in revenue and CNY33.379 billion (approx. US$5 billion) in net profit attributable to the parent company. Remarkably, its profit in a single quarter eclipsed its cumulative net earnings from prior years. This has turned YMTC's CNY33 billion (approx. US$4.9 billion) fundraising plan into one of the most anticipated semiconductor IPO in China since ChangXin Memory Technologies (CXMT).

Taiwanese petrochemical maker Grand Pacific Petrochemical (GPPC) announced it has joined Japanese industrial gas manufacturer Air Water (AWI) in a capital increase for semiconductor specialty gas producer Hong-Kuang Hi-Tech (HK Hi-Tech). The move marks their entry into Taiwan's semiconductor specialty materials market as part of a joint Taiwan-Japan effort to develop the local industry.

The five largest cloud operators spent US$181.5 billion on property and equipment in the June quarter, 87% more than a year earlier. Nvidia's hyperscale revenue was US$48.7 billion — 26.8 cents of every dollar they spent, down from a peak of 33.2 cents three quarters ago.

Doosan Corporation's Electro-Materials BG has developed copper-clad laminate (CCL) technology that reduces the prepreg (PPG) insulating layer used in memory semiconductor package substrates from about 18µm to 13µm, and is preparing the material for mass production as thinner package designs raise the technical demands on substrate materials.

Kioxia Holdings is weighing construction of a new manufacturing building at its Kitakami plant in northern Japan, as rising NAND demand pushes the memory maker toward its next phase of capacity expansion beyond 2029.

South Korean memory rivals Samsung Electronics and SK Hynix are taking fundamentally different routes to solve HBM performance bottlenecks. While Samsung frames processing-in-memory (PIM) as its core solution, SK Hynix is prioritizing packaging interconnect innovation to streamline data movement.

Nvidia said in its fiscal 2027 second-quarter results on the 27th that it sold a small number of H200 artificial intelligence (AI) processors to Chinese customers last quarter, marking its first renewed AI chip sales to China since shipping about US$60 million of H20 chips in early 2025. But shipments fell short of the total approved by US President Donald Trump and licenses cleared by the US government in January, amid opposition from Beijing.

Nvidia closed its second fiscal quarter with US$31.6 billion of inventory, but the more consequential change is what that inventory now consists of. Work in process and raw materials together account for 78% of the balance, against 42% a year earlier. The mix shows a company committing to scarce memory well ahead of the Vera Rubin ramp, and absorbing the cost several quarters before customer price increases land.

Nvidia's data center business now splits almost evenly between hyperscalers and everyone else — US$48.7 billion against US$40.3 billion in the July quarter. The obvious reading is that the customer base is broadening. The sequence says something more specific.

CXMT is narrowing the performance gap with SK Hynix in high-speed DDR5 memory, with new benchmark tests showing CXMT chips running stably at 8,000 MT/s and delivering productivity and gaming performance within about 1% of SK Hynix's latest M-die.