AI Innovation Soars as Nvidia Announces 2026 Chip Platform

Nvidia Corp.'s chief executive officer, Jensen Huang, said the firm intends to improve its AI Innovation every year. In 2025, a Blackwell Ultra chip will be released, and in 2026, a next-generation platform known as Rubin will be developed. New tools and software models were revealed by the business, which is now most recognized for its artificial intelligence data center systems, on the eve of the Computex trade expo in Taiwan. The CEO said that Nvidia sees the rise of generative AI as a new industrial revolution and that the company plans to play a significant role in the shift from technology to personal computers during a keynote address at National Taiwan University.

The company that became the world's most valuable chipmaker, Nvidia, has significantly profited from a massive increase in AI spending. Its present goal, nevertheless, is to expand its clientele beyond the handful of significant cloud computing firms that provide the lion's share of its income. Huang anticipates more businesses and government organizations will use AI as part of the development, including everyone from pharmaceutical firms to shipbuilders. Back in the exact location, he tackled topics from a year before, such as the notion that those without AI skills will be left behind.

“We are seeing computation inflation,” Huang said on Sunday. Traditional computing techniques are unable to handle the exponential growth of data that has to be processed, and Huang claimed that the only way we can reduce prices is by utilizing Nvidia's accelerated computing approach. He highlighted how Nvidia's technology resulted in 98% cost savings and 97% less energy use, saying that it constituted “CEO math, which is not accurate, but it is correct.”

Right after the news, the stock prices of Taiwan Semiconductor Manufacturing Co. and other suppliers increased. TSMC's stock increased by as much as 3.9%, and Wistron Corp.'s stock followed with an increase of 4%.

Huang indicated that the next generation of high-bandwidth memory, known as HBM4, will be used in the next Rubin AI platform. HBM4 has become a manufacturing bottleneck for AI accelerators, with leader SK Hynix Inc. nearly sold out through 2025. Other than that, he did not provide specifics on the following goods, which will come after Blackwell.

“I think teasing out Rubin and Rubin Ultra was extremely clever and is indicative of its commitment to a year-over-year refresh cycle,” said Dan Newman, CEO and chief analyst at Futurum Group. “What I feel he hammered home most clearly is the cadence of innovation and the company’s relentless pursuit of maximizing the limit of technology, including software, process, packaging, and partnerships to protect and expand its moat and market position.”

Nvidia's history of selling gaming cards for desktop PCs comes into play as computer manufacturers strive to include more artificial intelligence (AI) features into their devices.

Microsoft Corp. and its hardware partners will showcase new laptops with AI advancements under the Copilot+ brand at Computex. Most of those upcoming gadgets are built on a new processor type from Qualcomm Inc., a rival to Nvidia, which will allow them to run longer between battery changes.

Although Nvidia says that adding a graphics card may considerably improve the devices' performance and allow new features for popular apps like gaming, the devices are still suitable for basic AI activities. The company claims that PC makers, such as Asustek Computer Inc., sell these devices.

Nvidia is providing software developers with tools and pre-trained AI models so they may add more new features to the PC. These models will manage complex activities like determining whether to process data locally or transfer it across the Internet to a data center.

Nvidia is announcing a new architecture for server systems that are based on its processors alone. Utilizing the MGX program, businesses like Dell Technologies Inc. and Hewlett-Packard Enterprise Co. may expedite the launch of enterprise- and government-grade products. With servers that pair their CPUs with Nvidia chips, competitors Advanced Micro Devices Inc. and Intel Corp. are also utilizing this design to their advantage.

The day following Huang's remarks, Lisa Su, the CEO of AMD, spoke at Computex and outlined her company's advancements in AI technology. In an effort to catch up to Nvidia in the rapidly expanding industry, AMD is accelerating the release of its AI chips.

Spectrum X for networking and Nvidia Inference Microservices, or NIM, which Huang dubbed "AI in a box," are two of Nvidia's previously announced solutions that are currently publicly accessible and being actively used, the firm claimed. Additionally, free access to the NIM products will be provided. A collection of intermediary tools and models known as microservices enables businesses to launch AI services more rapidly without having to worry about the underlying technology. Businesses that use them are then required to pay Nvidia a use charge.

Huang also advocated the use of digital twins in an Ominverse virtual environment, as described by Nvidia. He used Earth 2, a digital clone of the planet Earth, to demonstrate the scope of what is conceivable by showing how it may be used to perform more intricate jobs like simulating complicated weather patterns. He mentioned how the tools are being used by Taiwan-based contract manufacturers like Hon Hai Precision Industry Co., popularly known as Foxconn, to improve factory operations and formulate better planning.