Global GPU Computing News【20260531】

1. Microsoft and NVIDIA to jointly launch Arm‑based PC chip

Towards the weekend, Microsoft and NVIDIA teased “a new era for personal computers” on social media. Reports say the two will unveil an Arm‑based PC chip (likely named N1 and N1X) at Microsoft Build and Computex next week. The chip combines NVIDIA’s low‑power GPU with MediaTek’s Arm CPU. The N1X variant is expected to feature CUDA cores comparable to a desktop RTX 5070, running Windows on Arm and directly challenging the x86 PC dominance of Intel and AMD.

2. NVIDIA Computex preview: new Rubin GPU and Vera CPU architecture

Next week’s Computex is expected to showcase NVIDIA’s next‑generation AI computing architecture, including the Rubin GPU and Vera CPU. According to Morgan Stanley, the Vera CPU may deliver up to 1.5× the performance of x86 alternatives. Analysts estimate shipments of 1.2 million and 4.2 million units for FY27E and FY28E respectively, representing a $20 billion market opportunity. NVIDIA also announced the start of its RTX Summer 2026 campaign, giving away RTX 50 series graphics cards and PCs.

3. Domestic AI chips receive highest national security certification for the first time

China’s Information Security Evaluation Center and the State Cryptography Administration’s Security Evaluation Center have, for the first time, included AI training/inference chips in their trusted computing evaluation system. Nine AI chips from seven Chinese companies (Huawei HiSilicon, Alibaba T‑Head, Biren Technology, Hygon, Iluvatar CoreX, M‑Thread, Moore Threads) all received the top Security Level I certification. Domestic AI chips now account for 41% of China’s AI server market, and analysts expect that share to reach about 76% by 2030.

4. Loongson’s first discrete GPU 9A1000 taped out, achieving full domestic production

On May 27, Loongson revealed that its first discrete GPU, the 9A1000, has been taped out on a domestic high‑autonomy process line, with test chips expected back soon. This marks a complete domestic闭环 (closed loop) from design to manufacturing. The 9A1000 targets entry‑level discrete graphics, with performance comparable to AMD’s RX550. Loongson also announced a planned fundraising of up to RMB 2.3 billion for next‑generation CPU and GPU R&D, with the mid‑to‑high end 9A2000 and 9A3000 already in planning.

5. Domestic GPU firm XianTI completes new financing, accelerates IPO

XianTI Computing Technology (Chongqing) completed a new funding round in May, led by Wise Road Capital and Junxin Investment, with participation from Guangzhou Yuegang Fund and existing investors. Funds will be used for next‑generation “Shennong” chip development, mass production, and software ecosystem building. The company plans to complete a shareholding reform within the year and accelerate its IPO. XianTI focuses on a “graphics rendering + AI inference” approach; its Tianjun series desktop GPUs approach the performance of NVIDIA’s 2070/3050 and have been in mass production since Q1 2026.

6. Alibaba DAMO Academy releases GPU‑based “heart of industrial software” solver

On May 28, Alibaba DAMO Academy released the GPU version of its MindOpt solver. Solvers are often called the “heart of industrial software”, essential for power grid scheduling, flight planning, and high‑end manufacturing. In tests on nearly 2,000 general benchmarks, Alibaba’s GPU solver improved large‑scale problem success rates by over 14% and average solution speed by 2.67× compared to mainstream competitors. It can also solve linear programming problems with hundreds of millions of variables.

7. Tencent’s self‑developed Canghai chip wins top honors at MSU video coding competition

Results from Moscow State University’s hardware video coding competition (May 26) show Tencent’s Canghai chip achieved first place across all core metrics (SSIM, PSNR, VMAF) for 30‑240fps ranges, outperforming AMD, NVIDIA, and Intel by over 30% in several tests. The first‑generation Canghai V1 chip has been deployed in over 100,000 units for Tencent Cloud, live streaming, and cloud gaming. The new V2 chip is already in full production with improved dual‑standard and full integration capabilities.

8. Private equity giants arrange $36 billion TPU financing for Anthropic to challenge NVIDIA

Apollo Global Management and Blackstone are arranging about $36 billion in debt financing for AI company Anthropic to purchase Google‑customized TPU chips. Broadcom will guarantee most of the loan, making it one of the largest private credit deals and the biggest chip‑financing debt transaction ever. Although NVIDIA’s GPUs still dominate the AI chip market, Google TPUs are becoming strong competitors. This massive financing will further boost TPU deployment in data centers.

9. NVIDIA invests over $6.5 billion in photonics technology in three months to tackle power bottlenecks

Since March, NVIDIA has committed at least $\mathrm{<span\; style="color:\#FFFFFF;">6.5</span>}\mathrm{<span\; style="color:\#FFFFFF;">b</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">v</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">r</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">p</span>}\mathrm{<span\; style="color:\#FFFFFF;">h</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">h</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">g</span>}\mathrm{<span\; style="color:\#FFFFFF;">y</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">m</span>}\mathrm{<span\; style="color:\#FFFFFF;">p</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}<span\; style="color:\#FFFFFF;">,</span>\mathrm{<span\; style="color:\#FFFFFF;">b</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">m</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">g</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">f</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">h</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">m</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">v</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">r</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">g</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">b</span>}\mathrm{<span\; style="color:\#FFFFFF;">u</span>}\mathrm{<span\; style="color:\#FFFFFF;">y</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">r</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">h</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">f</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">d</span>}\mathrm{<span\; style="color:\#FFFFFF;">.</span>}\mathrm{<span\; style="color:\#FFFFFF;">I</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">v</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">m</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">s</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">u</span>}\mathrm{<span\; style="color:\#FFFFFF;">d</span>}\mathrm{<span\; style="color:\#FFFFFF;">e</span>}$2 billion each to Lumentum, Coherent, and Marvell, $\mathrm{<span\; style="color:\#FFFFFF;">500</span>}\mathrm{<span\; style="color:\#FFFFFF;">m</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">l</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">C</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">r</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">g</span>}<span\; style="color:\#FFFFFF;">,</span>\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">d</span>}\mathrm{<span\; style="color:\#FFFFFF;">p</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">r</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">c</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">p</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}\mathrm{<span\; style="color:\#FFFFFF;">t</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">o</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">i</span>}\mathrm{<span\; style="color:\#FFFFFF;">n</span>}\mathrm{<span\; style="color:\#FFFFFF;">a</span>}$500 million Series E for optical startup Ayer Labs. Photonics, which uses light instead of electrical signals for data transmission, is seen as a key path to overcoming power and scalability limits in AI infrastructure. Analysts expect large‑scale commercial adoption around 2028, but NVIDIA’s forward‑looking moves are already underway.

10. Huawei reportedly investing heavily in AI computing infrastructure, Ascend chips nearing one million shipments

As domestic AI chip share continues to rise, Huawei’s Ascend series stands out. Data shows that in the second half of 2025, domestic AI chips surpassed a 40% combined market share, with Huawei alone shipping nearly one million units. Both the Ascend 910 flagship AI training chip and the Ascend 310 high‑efficiency AI inference processor are on the national Security Level I certification list. Huawei’s Ascend ecosystem is being rapidly built, becoming a key component of China’s AI computing base. Analysts estimate Huawei’s AI processor revenue could exceed $12 billion in 2026.