Corning’s $6 billion agreement with Meta, announced in 2023, was meant to be the crown jewel of its optical networking resurgence. But on May 07, 2026, CEO Wendell Weeks told Jim Cramer the real story is far bigger — and it’s not with Meta.
Key Takeaways
- Corning has signed two separate deals with unnamed hyperscalers that Weeks described as larger than the $6 billion Meta pact
- The deals center on optical connectivity infrastructure critical to AI data centers, not consumer glass
- Weeks emphasized Corning’s manufacturing scale and proprietary glass science as key differentiators
- Neither hyperscaler has been named, but demand signals point to Azure and Google Cloud as likely candidates
- Corning’s stock rose 7.3% by midday trading on May 07, its sharpest single-day gain in 18 months
Not the Glass You Think
When most people hear “Corning,” they think of Gorilla Glass — the scratch-resistant panels on iPhones and tablets. That business still matters. But the money — the real momentum — is in a quiet, unglamorous product: optical fiber and interconnects.
These aren’t the same cables laid by telecoms in the 1990s. These are single-mode, ultra-low-loss fibers engineered to move data at 800 Gbps and beyond between AI accelerators inside data centers. A single rack of GPUs can generate more internal traffic than an entire corporate network did five years ago. That traffic doesn’t go over air or copper. It goes over glass.
And Corning makes the purest, most precisely engineered glass on the planet. In fact, the company has developed new manufacturing techniques that enable it to produce glass with imperfections as small as 1 nanometer. This level of precision is crucial for maintaining signal quality and minimizing signal loss over long distances.
The Hidden Layer of AI Scaling
Everyone focuses on chips. The H100s. The Blackwell platforms. The custom TPUs. But none of that matters if data can’t move fast enough between components. That’s where optical interconnects come in. They’re the plumbing — the capillaries — of AI infrastructure.
Weeks didn’t use jargon. He put it plainly: “If you can’t move data from one server to another in under 100 nanoseconds, your AI cluster doesn’t scale. You’re bottlenecked. We solve that.”
That bottleneck is real. At scale, electrons in copper traces heat up, distort signals, and limit bandwidth. Light in glass doesn’t. Corning’s specialty is drawing glass into fibers so pure you can see through a 6-mile column of it. That purity means lower signal loss, fewer repeaters, and higher data density.
In fact, Corning’s fiber optic cables can transmit data at speeds of up to 10 Gbps per wavelength, which is faster than most modern data centers can currently handle. This means that as data centers continue to grow and expand, Corning’s optical interconnects will be crucial for maintaining performance and scalability.
How Hyperscalers Buy Differently Now
Five years ago, hyperscalers bought optical components in pieces — transceivers from one vendor, cables from another, switches from a third. They assembled the network stack themselves. Now, they’re buying integrated solutions. Full interconnect systems. Plug-and-play optical backbones.
That shift plays directly into Corning’s strength. The company doesn’t just make parts. It designs, manufactures, and qualifies entire optical architectures. And it does it at scale no competitor can match.
Consider this: Corning’s factory in Wilmington, North Carolina, produces 1.2 million kilometers of fiber per year. That’s enough to wrap around the Earth 30 times. And according to Weeks, it’s not enough to meet current demand.
This manufacturing scale is crucial for hyperscalers, who need to deploy optical interconnects at a massive scale to support their AI workloads. Corning’s ability to produce high-quality optical fibers and interconnects at scale makes it an attractive partner for hyperscalers who need to build out their AI infrastructure.
Larger Than $6 Billion — But With Who?
The Meta deal, signed in 2023, was structured as a multi-year, $6 billion commitment to Corning’s optical products. At the time, it was the largest non-M&A agreement in Corning’s 170-year history. Weeks called it a “watershed.”
But on May 07, he told Cramer: “We have two other agreements — with hyperscalers you know — that are individually larger than that Meta deal.”
“We’re not talking about billions. We’re talking about multiples of that.” — Wendell Weeks, CEO of Corning, on CNBC’s Mad Money, May 07, 2026
He wouldn’t name them. But the market reacted instantly.
By 10:14 a.m. ET, Corning stock had jumped 5.1%. By noon, it was up 7.3%, closing the day at $44.82 — a level not seen since 2021.
Speculation immediately turned to Microsoft and Google. Amazon has its own internal optical initiatives and has historically relied more on vertical integration. Microsoft, expanding Azure AI regions globally, has been vocal about infrastructure constraints. Google, meanwhile, is deploying TPU v6 pods at scale — and those need dense, low-latency interconnects.
Why Scale Beats Innovation in This Race
Startups like Ayar Labs and Celestial AI are building silicon photonics chips that could one day replace discrete optical modules. Their tech is elegant. It integrates lasers and modulators directly onto silicon. But they can’t manufacture at volume — not yet.
Corning can. It has 30,000 employees in manufacturing and R&D. It spends $1.1 billion annually on materials science. It owns the entire supply chain — from raw silica to finished cable reels.
When a hyperscaler needs 500,000 fiber trunks delivered to a new data center campus in Iowa by Q3, they don’t want a prototype. They want reliability. They want volume. They want Corning.
The Competitive Landscape
While Corning is a leader in the optical interconnect market, it’s not the only player. Other companies, such as Infinera, Arista Networks, and Cisco Systems, also offer optical interconnect solutions. However, Corning’s scale and manufacturing capabilities give it a significant advantage in this space.
According to a recent report by the market research firm, Dell’Oro Group, Corning held a 34% market share of the optical interconnect market in 2022, followed by Infinera at 23% and Arista Networks at 17%. This market share is expected to continue to grow as hyperscalers increasingly rely on optical interconnects to support their AI workloads.
Corning’s dominance in the optical interconnect market is driven by its ability to produce high-quality optical fibers and interconnects at scale. The company’s manufacturing capabilities, combined with its strong research and development efforts, have enabled it to stay ahead of the competition and meet the growing demand for optical interconnects in the AI market.
What This Means For You
If you’re building AI systems — whether at a startup or inside a large cloud team — understand that the physical layer still matters. Your model’s performance isn’t just about FLOPs and parameters. It’s about latency between nodes, thermal load from interconnects, and packet loss in the fabric. These deals suggest hyperscalers are betting that optical networking will define the next ceiling of AI scale.
For infrastructure engineers and developers, this means two things: First, learn how optical interconnects affect distributed training performance. Second, watch Corning’s product roadmap. When they release a new fiber spec or connector type, it’s a leading indicator of what’s coming in cloud AI architecture. This isn’t just hardware news. It’s a signal about where the bottlenecks are moving.
So if Corning is signing multi-billion-dollar deals for optical infrastructure — deals larger than its landmark Meta agreement — then the real question isn’t who the hyperscalers are. It’s how much AI capacity the world is actually building, and whether even Corning can keep up.
Key Questions Remaining
As Corning continues to grow its optical interconnect business, there are several key questions that remain unanswered. For example, how will Corning’s manufacturing capabilities be stretched to meet the growing demand for optical interconnects? Will the company be able to maintain its market share in the face of increasing competition from other vendors?
Another question is how Corning’s optical interconnects will be integrated into the broader AI infrastructure landscape. Will they be used as a standalone component, or will they be integrated into existing systems and architectures?
Finally, what implications will Corning’s dominance in the optical interconnect market have for the broader cloud and AI ecosystem? Will it drive further innovation and investment in optical networking technologies, or will it lead to a consolidation of the market and a reduction in competition?
Sources: CNBC Tech, original report
