On the afternoon of May 6, NVIDIA announced an investment. The amount of money involved was not particularly large, $500 million. However, the contract stipulated that an additional $3.2 billion could be added in the future. Corning's stock price rose by 14% that day.

What's more intriguing is the structure of this transaction. Among the 18 million stock warrants that Corning issued to NVIDIA, 3 million of the shares have an exercise price of $0.0001. This means that these 3 million shares were almost gifted to Corning. On the same afternoon, at an investor event in New York, Corning raised its revenue growth target for 2030 to $40 billion.

However, these are not the most unusual aspects of Corning in recent months. In the first-quarter earnings report of this "iPhone screen glass supplier," it was disclosed that in the past few months, Corning had signed multi-year contracts worth $6 billion each with two unnamed companies. The reason for using "again" is that Corning had just recently signed a similar-sized contract with Meta.

A count would reveal that over the past four months, there have been at least four AI mega-deals worth tens of billions of dollars, all concentrated on this 174-year-old glass company. In the last six months, Corning's stock price has risen by 140%, and compared to two years ago, it has increased nearly fivefold.

From Selling Phone Glass to AI Factory Hotspot

If you are reading this article on your phone, the screen most likely has a piece of Corning-produced glass. Starting from Apple's first iPhone in 2007, Corning's Gorilla Glass has almost become the default option for high-end smartphone screens worldwide. However, being a "phone glass supplier" is just one aspect of Corning and not the most lucrative one.

Corning's Gorilla Glass production line, Image Source: Apple

Founded in 1851, the company created the glass shell for Edison's first incandescent light bulb and independently invented low-loss optical fiber in the 1970s, revolutionizing the entire modern fiber optics industry. The 2007 iPhone glass marked its third significant business pivot. Today, Corning is undergoing its fourth transformation, with optical communications becoming its true business driver.

Corning's optical communication business has a history of more than 50 years, but the customer structure of this business underwent a complete reversal in the past two years.

For a long time, Corning's optical fiber was mainly sold to telecommunications operators such as AT&T and Verizon. These companies used it to deploy fiber-to-the-home and build 4G and 5G base stations. In 2009, Corning introduced a data center cabling solution called EDGE, officially adding data center operators to its customer list. Over the past decade, with the explosion of mobile internet, the proliferation of cloud services, and the explosive growth of remote work during the pandemic, Corning's optical communication business has been steadily rising but has never been the revenue driver.

In November 2022, OpenAI introduced ChatGPT to the public. From that moment, data centers worldwide began redesigning their physical infrastructure for this new computing task required for AI training. The fiber density needed for AI training has never been seen in any previous era.

The earliest signs appeared in August 2024. A U.S. telecommunications operator named Lumen placed a one-time order for 10% of Corning's global optical fiber capacity, two years in a row. This was the earliest public signal of Corning's business transitioning to the AI field.

By early 2026, the four $6 billion-level contracts mentioned earlier exploded in concentration. Corning had been working with data center operators for 15 years, but the "secondary customers" turned into the "absolute main force" only in the past 24 months.

The direct impact of this customer reversal was reflected in Corning's financial reports. Corning's full-year revenue in 2023 fell by 11% year-on-year, hitting a low point for the industry. However, in 2025, full-year revenue surged to $15.6 billion, a 19% year-on-year increase. In the first quarter of this year, its revenue increased by another 18% year-on-year. The most substantial growth came from the optical communication business, which grew by 35% for the full year. The proportion of optical communication in total revenue increased from 30% in 2020 to 37% in 2025. The absolute value change is more intuitive, growing from $2 billion five years ago to $6.3 billion in 2025, more than tripling.

This transformation from a "secondary business" to the "flagship" was not accidental; it was part of a growth plan led by the company's CEO Wendell Weeks. This plan has an internal code name, called Springboard, which literally means "diving board."

Two years ago, Corning was still seen by Wall Street analysts as a "boring glass manufacturer," classified as a mature, low-growth dividend stock. However, three years after the implementation of the Springboard plan, Corning's stock price soared from just over $30 in early 2024 to $162, a fivefold increase in two years, with a direct 140% surge in the past six months. The glass factory transformed into the "neural system of the AI revolution."

Springboard was first announced in September 2024. It started with the annualized revenue level in the fourth quarter of 2023, around $13 billion. The initial goal was to increase the annualized revenue by over $3 billion by the end of 2026, with an overall operating profit margin of 20%.

However, over the next year and a half, this target was raised three times in a row, all the way up to $65 billion, pushing the annualized revenue by the end of 2026 to a $200 billion scale. After NVIDIA's investment in Corning on May 6, the company directly raised its internal revenue target for 2030 to $400 billion. At the same time, by the fourth quarter of 2025, Corning had already achieved the 20% profit margin target a year ahead of schedule.

The key to Springboard's plan lies in "premium pricing." The company's sales grew by 18%, but earnings per share increased by 46%, with profit growth 2.5 times that of sales growth. From a business perspective, Corning mainly did three specific things:

The first was to raise prices for existing businesses. Corning's display glass had been a mature business with no growth for many years. However, by the end of 2024, Corning raised prices on this line by over 10% while locking in the yen exchange rate until 2030. As a result, even in an environment of yen depreciation, this line consistently contributed $900 million to $950 million in annual net profit by 2030, maintaining a net profit margin of 25%.

The second was the upgrade of optical communication products. Throughout 2025, optical communication sales increased by 35%, but net profit increased by 71%. This means that not only did optical communication sell more, but it also earned more per optical fiber.

The third was to utilize idle capacity. Rather than constructing new factories on a large scale, Corning restarted previously idle capacity during past cyclical troughs, raising the company's overall gross margin from 33% in 2024 to 36% in 2025.

Of course, the ability to increase prices is because there are customers willing to pay. The ability to earn more through product upgrades is because there are customers willing to pay more for upgraded products. The reason why Springboard allowed Corning's profit growth rate to outpace revenue is essentially because a group of customers willing to pay a premium is now part of its customer base.

Everyone is Snatching Up Fiber Optics

The AGI competition and order demand have made every data center operator incredibly anxious about time.

The core business of cloud giants has always been "renting IT to enterprises." Companies like Netflix, Airbnb, and Uber, which emerged with the mobile internet, mostly have "north-south" traffic. A user opens an app from outside, and the request is sent to a cloud server, which then returns the data. While servers also communicate with each other occasionally, the volume and frequency are not high. This kind of network architecture does not have strict requirements for the underlying physical infrastructure: Ethernet is sufficient, copper cables are sufficient, regular fiber optics are sufficient. Cloud giants have been using this architecture for over a decade, operating smoothly, stably, and profitably.

However, the rules of the game began to change with the emergence of ChatGPT.

Over the next few years, almost all cloud giants started to get into training themselves. Microsoft is the largest provider of compute power for OpenAI, AWS is deeply integrated with Anthropic, and Alibaba focuses on general intelligence. The core business of cloud giants has shifted from "renting IT to enterprises" to "training AI for the world."

But the ripple effects of this transformation at the physical infrastructure level have surpassed all accumulated knowledge of the past 20 years.

The traffic characteristic of AI training is "east-west." Training a large model may require tens of thousands of GPUs to communicate with each other simultaneously, synchronizing the gradients calculated by each other. If any line is slow, the entire training phase has to wait for it, and tens of thousands of GPUs together become "cars stuck at an intersection." Therefore, the east-west traffic has latency and bandwidth requirements dozens of times higher than the past north-south traffic.

Prior to this, most high-speed connections within data centers were copper cables. Copper, being cheap, easy to install, and performing well, has always been the default option for data centers. However, the geometric structure of AI training clusters is precisely what copper cables dislike the most. Tens of thousands of GPUs spread across dozens of racks, each a dozen meters apart, cannot be connected by copper. On the other hand, fiber optics have no distance limit in this regard.

Overnight, the previously sufficient sparse network became insufficient. Cloud giants had to re-lay fiber optics in a denser manner than ever before.

The scale of this re-laying is already reflected in their capital expenditures. By 2026, the total capital expenditure of the world's six cloud giants is expected to exceed $600 billion. The number of ultra-large-scale data centers in operation globally will reach 1,297, nearly three times the number in early 2018. In just the year 2026 alone, the projected number of new data centers is expected to exceed 150, with corresponding AI infrastructure construction spending exceeding $400 billion.

Market research firms estimate that the total demand for fiber optics in AI clusters is 10 to 100 times that of traditional cloud services. This is the fundamental reason why Corning is now able to secure four $6 billion deals.

Between data centers and within data centers, and even between racks, all fiber optic cables have to go through something called a duct. It is usually a 2-inch to 4-inch diameter tube made of plastic or metal, either buried underground or running overhead on racks. Once these ducts are laid, it is difficult to add more. Adding a new duct between cities means reapplying for right-of-way, digging up the road again, taking months if not years. Adding a duct in an already operational data center means downtime for construction and months of work.

Fiber optic duct about to be buried underground, Image Source: The Network

In the past couple of years, Corning has been specifically working on enabling existing ducts to accommodate more fiber without adding new ducts.

In addition to making the fiber itself thinner, Corning has also changed the way the fibers are arranged from a loose "spaghetti" style to a flat ribbon that can be rolled up when not in use, and then laid flat when needed, allowing for a tighter arrangement inside the cable. While a 2-inch duct could originally hold only about 1,000 fibers, Corning's new design can accommodate over 3,000 fibers, doubling the capacity. If a 4-inch duct is used with 6 of these ribbon cables laid side by side, it can hold over 20,000 fibers, which is more than 6 times the capacity of traditional designs.

Corning's rollable optical fiber, Image Source: Corning

Not only can more fibers fit, but the process of termination is also more efficient. A cable with 3,456 fibers would traditionally take over 200 labor hours to terminate individually, but with Corning's ribbon design, this can be reduced to under 40 hours, reducing cable preparation time by 30%. It is important to note that the U.S. already faces a shortage of fiber optic engineers.

During the construction of a large-scale AI facility, each month of delay translates to significant GPU depreciation and delayed training tasks, amounting to billions on the balance sheet. A product that can shave off several months and millions of dollars in engineering costs is well worth the 30% to 70% premium spent on fiber optics.

Jensen Huang's "Unprecedented Scale"

On May 8, NVIDIA CEO Jensen Huang reiterated in an interview that the next generation of AI infrastructure requires a large amount of optical connectivity, as copper wires are no longer sufficient. He also mentioned that NVIDIA aims to expand the application of optical technology at an unprecedented scale.

In the details of the recent investment transaction with Corning, we can indeed see this "unprecedented scale." Out of the 18 million shares, 3 million were granted for free. This structure is very rare in NVIDIA's ecosystem investments over the past year, indicating that NVIDIA obtained a significant equity exposure to Corning without needing to use cash upfront. This resembles more of a signature fee for a long-term partnership agreement.

Corning is not the only chess piece NVIDIA has placed its bet on. Since September last year, NVIDIA has entered a new investment rhythm. Firstly, the scale has increased, and secondly, the structure has begun to frequently utilize financial instruments such as "frameworks," "options," and "prepaid warrants," locking in commitments first before gradually redeeming them. In addition to the $100 billion investment framework for OpenAI, NVIDIA has consecutively made tens to hundreds of billions of dollars in follow-up investments in AI infrastructure companies such as Anthropic, Intel, and CoreWeave.

One of the most easily overlooked aspects is its investment in optical communication. In addition to Corning, NVIDIA has also invested $2 billion each in Lumentum and Coherent, two of the world's largest optical device companies. Including Corning's initial $500 million plus $3.2 billion in options, NVIDIA has invested approximately $7.7 billion in the optical communication sector alone.

If you were to put this investment table on a chart, you would see that they are essentially a blueprint for constructing an AI factory: computing power, networking, optics, power, cooling, software, customers, models—each layer has at least one key supplier locked in by NVIDIA. At this year's GTC conference, NVIDIA integrated this entire stack into a publicly available design, releasing a hardware reference architecture called Vera Rubin DSX and a digital twinning solution called Omniverse DSX Blueprint. The whole package is essentially the "blueprint for an AI factory."

An AI factory with a gigawatt-scale (power consumption equivalent to 1 million households) takes 18 to 24 months from planning to production, involving coordination with over 100 vendors. In the past, this was done by data center operators, each having to redo interface verification. However, NVIDIA's Omniverse DSX has systematized this process—products from all partners have already undergone validation in NVIDIA's digital twin, aligning parameters, standardizing interfaces, so cloud giants can simply purchase according to NVIDIA's blueprints.

Jensen Huang Unveils AI Factory Blueprint Platform at 2026 GTC Conference, Image Source: NVIDIA

This marks a crucial step for NVIDIA, transitioning from a chip company to an "AI Factory General Contractor." With increased integration and expanded gross margin, even if AMD or Broadcom were to produce GPUs with equivalent performance tomorrow, replicating this supply chain coordination from chip to fiber optics to power grid would take several more years at least.

Therefore, the true significance of NVIDIA's $32 billion deal with Corning lies in the fact that within its own AI factory blueprint, it has identified a key player to "localize optical communication production capacity." Of course, currently, NVIDIA is the only one capable of drawing up this blueprint.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia