Subscribe to Our Newsletter

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks
NVIDIA unveils AI computing systems, expands Taiwan operations
Philip Lee, Pickool

NVIDIA unveils AI computing systems, expands Taiwan operations

NVIDIA announces new Grace Blackwell GB300 systems, partnerships with Taiwanese manufacturers, and plans for a local AI supercomputer at Taipei industry event.

Philip Lee profile image
by Philip Lee

Taipei, Taiwan—NVIDIA Corp (NVDA.O) announced new computing systems and partnerships on Monday as the chip designer seeks to expand its presence in the artificial intelligence infrastructure market.

At an industry event in Taipei, CEO Jensen Huang presented updated versions of the company's high-performance computing products alongside new offerings aimed at personal and enterprise users.

The company plans to release its Grace Blackwell GB300 systems in the third quarter. 

It claims they will offer 50% more processing power for inference tasks and memory capacity than current models.

"These AI data centers are, in fact, AI factories," Huang told attendees. "You apply energy to it, and it produces something incredibly valuable."

NVIDIA also introduced a new architecture called NVLink Fusion, which allows third-party chip designers to integrate with NVIDIA's interconnect technology. 

According to the company, AlChip, Astero Labs, Marvel, and MediaTek have agreed to participate as partners.

In a local initiative, NVIDIA announced plans to build an AI supercomputer in Taiwan in collaboration with Foxconn, TSMC, and the Taiwanese government to serve researchers and businesses in the region.

For individual developers, NVIDIA revealed the DGX Spark, a smaller-scale computing system that the company says delivers one petaflop of performance. 

The system will be available through hardware partners, including Dell, HPE, Asus, MSI, Gigabyte, and Lenovo.

A higher-powered DGX Station model was also announced, which NVIDIA claims can deliver up to 20 petaflops of computing power through a standard wall outlet.

The company presented RTX PRO Servers for enterprise customers, which it says are now in production.

NVIDIA also announced several robotics-related developments, including a physics simulation engine developed with DeepMind and Disney, and new software tools for humanoid robot development.

The event concluded with the announcement of a new NVIDIA office in Taiwan.

Philip Lee profile image
by Philip Lee

Subscribe to The Pickool

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks

Read More