NVIDIA has unveiled its latest suite of technologies aimed at sustaining the surge in demand for AI computing power and solidifying its central role in the industry.
اضافة اعلان
On Monday, NVIDIA CEO Jensen Huang kicked off Computex — Asia’s largest electronics trade show, held in Taiwan — by promoting the company’s new products and strengthening ties with a region that plays a vital role in the tech supply chain. This comes as NVIDIA's stock rallies again following a deal-making tour in the Middle East as part of a business delegation led by former President Donald Trump.
Returning to his homeland, Taiwan, Huang introduced updates to NVIDIA’s lineup of accelerator chips — a cornerstone of developing and operating AI software and services. The primary goal is to broaden NVIDIA's product reach and remove barriers to AI adoption across more industries and countries.
Speaking about the region, Huang stated:
"When new markets must be built, they must be built from here — the heart of the computing ecosystem."
New Chips and Computing Systems
Huang announced that the next-generation GB 300 systems will launch in Q3 of this year. These will mark a significant upgrade to NVIDIA’s high-end Grace Blackwell AI systems, which are already being deployed by major cloud service providers.
NVIDIA also introduced a new version of its fully integrated data center computers. Through its NVLink Fusion products, customers will now have the option to pair their own CPUs with NVIDIA's AI chips — or use NVIDIA CPUs with third-party AI accelerators.
Openness in NVIDIA’s Designs
Until now, NVIDIA systems have only operated using its own components. However, this new openness in design — including vital connectivity components that ensure fast communication between processors and accelerators — will offer data center clients greater flexibility and introduce a measure of competition, all while keeping NVIDIA’s technology at the core of the ecosystem.
Major clients like Microsoft and Amazon are working on designing their own processors and accelerators, potentially challenging NVIDIA’s dominant role in data centers.
Huang also revealed that companies such as MediaTek, Marvell Technology, and Alchip Technologies plan to develop custom AI chips compatible with NVIDIA-based devices. Additionally, Qualcomm and Fujitsu are set to produce custom processors that will operate with NVIDIA accelerators in computers.
