NVIDIA has been a major beneficiary of the AI infrastructure wave since 2023; a $10,000 position bought at the start of 2023 would now be worth more than $131,000. Investors scouting alternatives point to AI supply-chain plays. Micron Technology has outperformed NVIDIA since 2023 — a $10,000 stake is now roughly $181,000, though most gains occurred in the past year. Micron, in a mid-stage expansion, says storage‑chip shortages driven by hyperscaler AI buildouts will persist beyond 2027, supporti

2026-07-09

NVIDIA has been a major beneficiary of the AI infrastructure wave since 2023; a $10,000 position bought at the start of 2023 would now be worth more than $131,000. Investors scouting alternatives point to AI supply-chain plays. Micron Technology has outperformed NVIDIA since 2023 — a $10,000 stake is now roughly $181,000, though most gains occurred in the past year. Micron, in a mid-stage expansion, says storage‑chip shortages driven by hyperscaler AI buildouts will persist beyond 2027, supporting higher prices, revenue and profit; the shares trade at a forward P/E of about 12.3x. Nebius Group, an early‑stage neocloud AI‑first cloud provider, reported Q1 revenue up 684% YoY and faces analyst projections of ~544% revenue growth in 2026 and ~234% in 2027; the company remains unprofitable and will likely require external equity or debt to fund further expansion.

其他消息
2026-07-09

At the A-share close, the Shanghai Composite rose 1.65%, the Shenzhen Composite Index gained 3.07%, ChiNext was up 4.49% and the STAR Market 50 Index jumped 8.41%. More than 2,400 stocks advanced as the semiconductor sector led gains, with memory chips, advanced packaging and lithography equipment among the top performers.

2026-07-09

China activated a national supercomputing internet core node: a single 100,000‑card full‑precision compute pool designed to support concurrent 100,000‑card training, ~10‑trillion‑parameter model full training, AI‑for‑Science and ultra‑large scientific engineering workloads, and to serve as a domestically developed inclusive large‑scale compute base. During trial it completed more than 70 tests at the 10k‑card scale. Using 45,000 cards it ran a 41.47 billion‑atom DFT‑precision simulation, setting