Continue Reading With Our 7-Day Free Trial
On Monday, Amazon announced a multiyear, multibillion-dollar agreement with Corning. Corning will supply optical fiber, cable and connectivity products for Amazon’s expanding US data center network. The deal will run for multiple years, create 1,000 jobs at Corning’s North Carolina manufacturing facilities, and carry a total value that neither company disclosed.
Corning’s stock opened up 9% on Monday morning and closed up roughly 5%. Amazon was up about 1%.
That asymmetry in the stock reactions is worth sitting with for a moment.
What Corning Actually Is
Most people know Corning from Gorilla Glass — the scratch-resistant coating on the front of every smartphone. That business still exists, but it is not the story anymore.
Corning has been in the fiber optic cable business since the 1970s. For most of that time it was a steady, unremarkable part of a large industrial conglomerate. Then AI data centers happened.
The architecture of an AI training cluster is fundamentally different from a traditional data center. Training a large language model requires thousands of GPUs working simultaneously as a single system. Those chips need to move enormous amounts of data between each other continuously. Copper cable, which was adequate for traditional computing, cannot carry the bandwidth at the distances involved without signal degradation. Optical fiber can. It carries more data, over longer distances, with less signal loss and lower power consumption.
Fiber is not glamorous. It is also not optional. And Corning is the dominant supplier.
Three Deals In Six Months
What happened Monday was not a one-off. It was the third major supply commitment Corning has received from a hyperscaler this year.
In January, Meta signed a deal worth up to $6 billion for Corning fiber through 2030. In May, Nvidia committed $3.2 billion and partnered with Corning to expand US-based optical connectivity manufacturing capacity tenfold and domestic fiber production by more than 50%. Now Amazon. Meta, Nvidia and Amazon have each independently concluded that they need to lock in Corning’s supply capacity before it runs out.
Corning CEO Wendell Weeks said as much earlier this year, describing fiber as the bottleneck the hyperscalers are now scrambling to solve. The fact that three of them have signed multi-year deals in the same six-month window suggests he was not exaggerating.
The financial impact is visible in the numbers. Corning’s Optical Communications revenue grew 36% year over year in Q1 2026. The stock has more than doubled in 2026 alone and is up roughly sixfold since the end of 2023.
What Amazon Is Signalling
Amazon is not just buying fiber. It is making a statement about the timeline and scale of its data center expansion.
AWS capital expenditure has been running well above consensus for four consecutive quarters. A multi-year supply agreement with a capacity-constrained supplier is what you do when you are confident your build schedule is not slowing down. Amazon is not hedging here — it is locking in.
For anyone tracking the AWS cloud revenue and margin story — which has been one of the more consistently underestimated narratives in large-cap tech — this deal adds another data point. Enterprise cloud migration is still in the early innings. Amazon is building for the next five years, not the next five quarters.
Where LENS Sits On All Of This
We launched the Baptista Research LENS Index — the Large-Cap Equity Narrative Strategy — 18 days ago. The index selects large-cap companies where the dominant market narrative is provably wrong and a specific catalyst will correct it. Every position enters with a published thesis, a price target, a named thesis break condition and a hard stop loss. Every call is documented before the trade.
Since inception on May 22, LENS is up 4.4% against the S&P 500’s 0.8% over the same period — 3.6 percentage points of alpha in under three weeks. We are not crowning ourselves yet. Three weeks is nothing. But the methodology is sound and it applies here.
So The Question Is: Does The Amazon-Corning Deal Change Anything For Our Model Portfolio?
Corning is up sixfold in two years. The market has fully priced the fiber bottleneck thesis. There is no narrative gap to exploit. Amazon trades at an elevated multiple on the back of AWS re-acceleration and the AI buildout consensus — both of which are already consensus.
Neither company carries a thesis that the market has meaningfully wrong right now. So neither Corning nor Amazon is in the LENS Index. Neither is on the Watch List.
But something connected to both of them has been on the Watch List since the day we launched — and the Amazon deal this week moved our timeline on it.
The fiber goes somewhere. The servers sit somewhere. Between the two sits a routing layer that almost nobody in this week’s coverage mentioned — and it is where we think the most durable margin in AI infrastructure actually sits.
Here is the full analysis, the company, and exactly what needs to happen before it moves from the Watch List into an active LENS position.
The Switching Layer
Every strand of fiber Corning manufactures terminates at a network switch. That switch is responsible for routing data between GPU racks — deciding, at line speed and with near-zero latency, where each packet of data needs to go next. At the scale of a modern AI training cluster — 10,000 to 100,000 GPUs — the switching architecture is not a commodity purchasing decision. It is a core infrastructure choice that determines whether the cluster performs as designed or spends half its time waiting for data.
Getting this wrong is expensive. Getting it right requires switching hardware that was purpose-built for AI workloads — high bandwidth, non-blocking, low-latency, with software that can manage traffic at a scale traditional enterprise networking vendors were never designed for.
One company controls more than 80% of this market. It has guided to $3.25 billion in AI-related revenue for 2026 alone. Its customers include every major hyperscaler — including the ones that just signed multi-year contracts with Corning.
That company is Arista Networks (ANET).
Why ANET Is On The Watch List, Not In The Index
Arista has been on the LENS Watch List since May 22 — the day we launched. Here is the entry exactly as it was written:
“80%+ AI ethernet switching share. $3.25B 2026 AI revenue target. Mandatory infrastructure between every GPU cluster. Trigger: Article published + 10% pullback from current levels.”
The thesis is straightforward. The dominant market narrative on Arista prices it as a beneficiary of AI capex — correct — but underestimates the structural irreplaceability of its position. Legacy vendors like Cisco and Juniper are not credible alternatives for hyperscaler-scale AI switching. Arista is not just winning business; it is the only option at this architecture level. That distinction matters enormously for long-term margin sustainability, and the market has not fully priced it.
The reason ANET is on the Watch List rather than in the index is valuation entry, not conviction. At current levels, the upside-to-downside ratio does not clear the LENS threshold. We want the thesis intact and a better price.
LENS Watch List — Arista Networks (ANET) Added: May 22, 2026 Tier on entry: BASE (3–5% allocation) 12-month target: $640 Thesis break condition: A named hyperscaler publicly moves to an in-house switching solution OR ANET guides AI revenue below $2.5B for 2026 Hard stop on entry: –30% from entry price Entry trigger: 10% pullback from current levels OR meaningful Q2 2026 earnings beat that re-rates the stock above our entry threshold
The Amazon-Corning deal does not trigger entry. What it does is reinforce that the data center buildout is real, multi-year and accelerating across every infrastructure layer — including the switching layer. Our conviction on the thesis is higher today than it was last week. Our entry price has not changed.
We are watching ANET’s next earnings print — and any commentary from Amazon or Microsoft on data center capex timelines — closely. If either event gives us the pullback or the re-rating we need, ANET moves from Watch List to active position before Q3.
Disclaimer: The Baptista Research Conviction Portfolio is a hypothetical analytical exercise for informational and educational purposes only. It does not constitute investment advice. Baptista Research is not a registered investment advisor. All investment decisions should be made in consultation with a qualified financial professional.
