CONNECT WITH US
May 21
How Nvidia plans to sell Vera CPUs: Four deployment models explained
Nvidia has unveiled a strategic expansion of its silicon portfolio, detailing a four-pronged commercial approach for its new Vera central processing unit. The chipmaker expects the processor to generate US$20 billion in standalone revenue, signaling a significant evolution in its hardware distribution strategy and a broader play for infrastructure dominance in the emerging agentic artificial intelligence market.
Khgears International said it would intensify a robotics push in 2026 by expanding from industrial robot gears into humanoid robot components and by seeking a strategic alliance with a Tier 1 supplier tied to one of Japan's four major robot families, executives said after a shareholders' meeting on May 21. The precision gearmaker reported that smart transmission products accounted for a record 12% of revenue in the first quarter of 2026 and expects smart transmissions to represent more than 10% of full-year revenue in 2026, up from about 7% in 2025.
Taobao is shifting its cross-border strategy from low-priced transactions to customer lifetime value by using artificial intelligence to surface latent demand and improve localized recommendations, the platform announced as it expanded into more than 200 markets and saw rapid growth in overseas downloads. Executives said the app ranked No. 1 in downloads across 16 overseas markets in April 2025 and that the overseas business had entered a phase of rapid growth.
A French consortium called AION said it would bid for EU AI infrastructure funding to build a large AI data center campus in France worth as much as EUR10 billion (US$11.6 billion). The group formed in 2025 and is pursuing the project as part of the EU's effort to boost domestic AI capacity and reduce reliance on US technology, executives said.
UiPath announced an on-premises Agentic AI deployment service and Automation Suite designed to help regulated industries adopt AI while maintaining data security and compliance. The vendor said the move, revealed in late May, targeted sectors including banking, financial services, insurance, healthcare, and the public sector in Taiwan and Greater China, where strict data residency and governance rules have limited cloud-only AI adoption.
On May 20, 2026, DIGITIMES held its Tech Forum, where senior analyst Jim Hsiao delivered a presentation, "AI server market outlook and trends amid the explosion of agent applications." In it, Hsiao explained that demand is rising sharply not only for x86 server CPUs from Intel and AMD, but also for Arm-based CPUs, which are expected to see explosive growth in 2026.
Nvidia has positioned its specialized LPX accelerator as a niche product designed specifically for low-latency, high-speed token generation rather than broad-market enterprise workloads. The hardware targets a narrow segment of service providers that operate premium, high-velocity token applications for select customer bases. While the architecture excels at rapid text decoding, its broader market adoption remains constrained by structural limitations in throughput and memory capacity.
At the recent Plug and Play Silicon Valley May Summit, the semiconductor and supply chain sessions highlighted a critical shift toward hardware-level security, energy efficiency for AI, and AI-driven quality engineering. Startups, including Enclave Semiconductor, Bedrock Semiconductor, and Lattice, presented solutions aimed at securing and optimizing the global electronics ecosystem.
Wayne Wang, senior director of commercialization at Robot Era, used the 5th Mobis Mobility Day in Sunnyvale to present a two-year-old Chinese robotics company that has already supplied hardware to nine of the world's top ten technology companies — including Apple, Nvidia, Amazon and Microsoft.
Nvidia has detailed a significant structural shift in its data center revenue reporting, introducing a new market segmentation that separates traditional hyperscale cloud providers from a rapidly expanding sector named ACIE, which stands for artificial intelligence clouds, industrial, and enterprise. While hyperscale platforms currently lead initial deployment, management projects that the ACIE segment will eventually outgrow them, owing to the massive scale of the global corporate economy.
Taiwan's export orders fell 4.0% in April compared to March levels, with orders from the US, China, Europe, Japan, and the ASEAN region all declining. According to data released on May 20 by the Ministry of Economic Affairs (MOEA) for April 2026, six of the seven major export categories saw month-on-month declines, including electronics and ICT products. However, the Ministry expects overall May orders to rebound from April, based on feedback from export-oriented manufacturers.
OpenAI is preparing to file for an initial public offering (IPO) in the coming weeks, a move that could sharpen its competitive edge in the AI race. While company executives had previously targeted a listing in the fourth quarter of this year or later, the accelerated filing is expected to substantially advance the offering's schedule. According to The Wall Street Journal, the initial paperwork could be submitted to regulators as early as Friday.