NVIDIA, a chipmaker at the center of the artificial intelligence boom, partners with a glass manufacturer corning Three new advanced manufacturing facilities in North Carolina and Texas focused on optical technology for the world’s most valuable semiconductor company.
The companies said in a joint press release Wednesday that the plants will create at least 3,000 jobs and increase Corning’s U.S. optical product manufacturing capacity tenfold.
Financial terms were not disclosed. Corning stock soared 17% on the news. Nvidia stock rose about 2%.
The deal gives Nvidia the right to invest up to $2.7 billion in Corning. Nvidia has warrants to buy up to 15 million shares of Corning common stock at a strike price of $180 per share, above Tuesday’s closing price of $162.10 but below the post-pop price.
In addition, NVIDIA has received pre-funded warrants to purchase up to 3 million shares of Corning common stock at an exercise price of $0.0001 per share for a total consideration of $500 million.
The multi-year agreement brings together two infrastructure players whose fortunes have soared since the announcement of OpenAI’s ChatGPT in 2022. This has sparked an explosion of investment in new processors and systems to power cutting-edge AI models and workloads. Although the companies have not disclosed details of the development, NVIDIA is likely preparing to replace copper with Corning’s optical glass fiber in its AI rack-scale systems, an integration known as co-packaged optics.
At Nvidia’s 2025 GTC conference, Nvidia CEO Jensen Huang said co-packaged optics are essential to building AI.
“What NVIDIA is doing is nothing short of extraordinary, not just for the future of AI, but for America’s advanced manufacturing workforce,” Corning CEO Wendell Weeks said in a press release.
Corning stock has risen more than 250% in the past year as of Tuesday’s close, helped by the 175-year-old company’s rapid pivot into the new economy. Meta announced in January that it would spend up to $6 billion to help lead customer Corning build an optical cable factory in Hickory, North Carolina, an expansion expected to create about 1,000 jobs.
NVIDIA CEO Jensen Huang speaks next to an NVIDIA Vera Rubin system at the NVIDIA GTC Global AI Conference on March 16, 2026 in San Jose, California, USA.
Fred Greavesreuter
Nvidia established itself in the AI market fairly early on, as its graphics processing units were key to developing large language models and giving big tech companies like Alphabet and Meta the ability to massively scale up their data centers. Nvidia’s stock has risen about 14 times over the past five years, but its gains have slowed in recent days as investors widened their bets on chipmaker Intel, memory provider Micron and a broader range of AI infrastructure companies that back Corning.
Analysts have long been looking forward to large-scale adoption of co-packaged optics by Nvidia, as the technology is expected to significantly increase data transfer speeds and reduce the energy needed for AI workloads.
Corning is best known for making all of the display glass for Apple’s iPhones, but optical communications remains the company’s largest and fastest-growing business. Since inventing fiber optics for long-distance communications in 1970, Corning has provided millions of miles of cable from every major company to connect racks in AI data centers.
copper replacement
By partnering with Nvidia, Corning is introducing fiberglass between the chips themselves, which could eventually replace 5,000 copper cables in rack-scale systems like Vera Rubin.
Fiber optic cables are small, bendable bundles of glass that can pass data as photons at much faster speeds and with less energy than those used in traditional copper wire.
“Moving photons uses five to 20 times less power than moving electrons,” Weeks said in a January interview with CNBC.
“It brings the photoconversion process very close to the computer chip,” said Vlad Galabov, head of enterprise infrastructure at research firm Omdia. “Because it only moves a few millimeters, it takes much less energy than moving an entire circuit board, so less power is wasted.”
Galabov added, “NVIDIA is accelerating innovation across the ecosystem.”
Fiber optics also has less signal loss than copper wire, which speeds up reliable communications and reduces the distance required between hundreds of thousands of GPUs in a data center.
“AI is driving the greatest infrastructure development of our time and represents a once-in-a-generation opportunity to reinvigorate America’s manufacturing and supply chains,” Nvidia’s Huang said in a press release. ”
“Together with Corning, we are inventing the future of computing with advanced optical technology and advancing our proud Made in America heritage while building the foundation for an AI infrastructure where intelligence moves at the speed of light.”
Nvidia released two network switches in 2025 that utilize similar technology and sit right next to its main AI chip. Competitors Broadcom and Marvell have announced similar products, and Intel is also developing co-packaged optical solutions.
Nvidia invested $4 billion in two companies, Coherent and Lumentum, in March. The company develops lasers and components that help convert data between optical and electrical signals transmitted through Corning’s fiber-optic cables.
“We’re working with all the different chip stakeholders on glass cores and how glass will be part of future semiconductor packages,” Weeks told CNBC during an exclusive factory tour in January.
“As the power problem becomes more and more important, fiber will inevitably move closer and closer to computing,” Weeks said. He added that as the number of GPUs in servers grows into the hundreds, “the distances will increase. As that distance increases, fiber becomes more economical and much more power efficient.”
Corning will host an Investor Day at the New York Stock Exchange on Wednesday, the day before the company rings the closing bell and celebrates its 175th anniversary.
Video: How Corning, a 175-year-old glass company, won a $6 billion AI infrastructure deal with Meta
