Earnings Highlights: Corning (GLW) | FY2026 Q3 — Optical Sales Up 44% to Half of Revenue as Customers Prepay to Fund Capacity

1. Executive Takeaway

First, optical is now the core of Corning. Optical Communications brought in $2.38B this quarter, 49% of GAAP revenue, up from about 40% a year ago. Enterprise (data center) sales grew 84%, with Gen AI products up more than 100%. Management stressed that this is clearly outpacing GPU shipments, meaning each GPU is consuming more fiber, not just that there are more GPUs.

Second, the customer-prepayment model has moved from glass and solar into optical. Customers pay cash up front to fund capacity, and Corning returns it as they purchase their committed volumes. The “customer deposits” line in operating cash flow added a net $639M over the first nine months. That lets Corning push 2026 capex to about $2B while adjusted free cash flow still reached $2.27B in the first nine months.

Third, the scale-up timeline holds: acceleration starts in 2028. CEO Wendell Weeks gave a blunt conversion: today’s scale-out architecture uses about 16 fibers per GPU; if scale-up went fully optical, that becomes 160, a tenfold jump. Scale-up today is 100% copper. On the noise about various component delays, his answer was that “this is just classic technology development” and the timeline stands.

2. Revenue and Financials

All figures below are GAAP, from the 8-K earnings release Corning filed with the SEC on October 27.

Metric	Q3 FY2026	YoY	QoQ
Revenue	$4.86B	+18.4%	+7.8%
Gross margin	37.3%	+0.2pp	—
Operating income	$911M (operating margin 18.8%)	+54.7%	—
Net income (attributable to Corning)	$680M	+58.1%	—
Diluted EPS	$0.78	+56.0%	—

Segment revenue:

Segment	Q3 revenue	YoY
Optical Communications	$2.38B	+44%
Glass Innovations	$1.51B	-3%
Automotive	$470M	+4%
Solar	$436M	+37%

▲ Corning FY2026 Q3 key financials (GAAP revenue / gross margin / operating margin / net income / EPS / YoY / Q4 guidance / segment revenue). (Image source: Simple Tech Trend original; data from the SEC 8-K earnings release (Exhibit 99) and Corning IR)

How to read the numbers:

• Operating income grew three times as fast as revenue. Revenue rose 18% YoY while operating income rose 55%. Gross margin added only 0.2 points, so the leverage came from expenses: SG&A fell from $624M to $563M. When an analyst asked in Q&A why corporate costs were about $50M lower YoY, the CFO said variable compensation ran above normal levels and there was no reallocation, adding that operating expenses will keep growing slower than sales.

• Mind the gap between GAAP and Core. The company’s own Core sales were $5.08B and Core EPS $0.91, about $230M of revenue above GAAP, mainly from the “adjustment for hedged exposures”: Corning restates its yen exposure at a ¥120 hedge rate, which hits Glass Innovations the most. The 19% growth and 40.5% gross margin quoted on the call are Core numbers.

• Optical has even steeper profit leverage. Segment net income was $532M, up 80% YoY, a 22.4% net margin. The CFO described the path as going from the “mid-teens” to the “low 20s,” saw room to keep improving, but is not giving a new target yet.

In our Corning FY2026 Q2 earnings highlights, Optical was $2.07B for the quarter; it added another $300M in one quarter, up 15% QoQ.

3. Optical Breakdown: Enterprise Up Fourfold, Carrier Pivots to DCI

Enterprise: growth comes from how much fiber each GPU uses, not how many GPUs ship. When the Springboard plan launched three years ago, Enterprise ran at about $1.4B annualized; it is now $6B, four times larger. The driver is new high-density products that fit 2 to 4 times more fiber in the same conduit. They were built for Gen AI clusters, but adoption is spreading faster than Corning itself expected.

Carrier: only 4% growth, but the growing piece is DCI. This quarter Corning signed long-term agreements with Zayo, Verizon and AT&T (on top of the earlier Lumen deal), spanning broadband and data center interconnect (DCI). Management had said DCI could reach $1B by the end of the decade; now it says it will “clearly exceed” that in both timing and size, without giving a new number yet.

Capacity is being allocated by priority. Asked by a Morgan Stanley analyst about Carrier and Enterprise competing for capacity, Weeks answered on two levels: traditional telecom customers also want the new high-density products for their DCI builds, so they compete for the same new lines; and allocation favors relationship tenure first, then “who is willing to help fund the capacity.” In other words, hyperscalers that can prepay will naturally get a bigger share.

Pricing is helping too. Weeks said overall prices are rising, and the value created by new products is shared with customers, so each unit of capacity yields more revenue. That is a signal you only see when supply is tight.

4. Customer-Funded Capacity: Springboard’s Funding Model Has Changed

This is the structural change most worth noting this quarter.

The traditional materials model: spend capex to build plants first, see revenue and cash flow years later, and carry the demand risk in between. Corning has now reversed that order:

1. Customers put up cash first to fund Corning’s expansion;

2. Corning ships against committed volumes;

3. Each time customers buy, Corning returns part of the deposit out of operating cash flow.

Demand risk sits with the party that knows best how much it needs: the customer.

The footprint in the numbers:

• Over the first nine months, “customer deposits and government incentives” added a net $639M in the cash flow statement;

• Capex in the same period was $1.30B, and adjusted free cash flow was $2.27B (versus $985M a year earlier);

• The CFO said that, excluding maintenance capex, most growth capex is funded by long-term customer agreements, and Corning usually gets the cash before it spends.

One detail to watch: in Q3 alone this line was a net outflow of $70M, meaning cash returned to customers exceeded new deposits. A Bank of America analyst asked directly whether this becomes a free cash flow headwind in 2027–2028; the CFO said the returns come out of incremental operating cash flow already earned, so they “smooth” rather than depress free cash flow. That claim needs to be checked against the next few quarters.

More context: Corning does more than sell fiber. As we broke down in ECOC 2026: fitting 1,152 fibers into a 1U, 102.4T CPO switch, the bottleneck in the CPO era moves from the chip to the fiber assembly line, exactly where Corning wants to sit.

5. Outlook: Q4 Guidance and the Scale-up Timeline

Q4 guidance (Core basis; the company gives no GAAP guidance):

• Core sales of $5.2–5.3B, up about 19% YoY;

• Core EPS of $0.96–1.00, up about 36% YoY.

At the $5.25B midpoint, the annualized run rate is about $21B, beating Springboard’s “$20B annualized by end of 2026” target a quarter early. The next targets stand: $30B by end of 2028 and $40B by end of 2030, roughly a 19% CAGR over 2027–2030.

Language check:

Topic	Management wording	Confidence
Q4 guidance	expect	High, with a numeric range
Scale-up goes optical, accelerating in 2028	continue to believe strongly / quite convicted	Medium-high, but no numbers
New optical margin target	not quite yet ready	Deliberately withheld
DCI above $1B	clearly exceed	Direction firm, size undefined
Solar profit improvement	slowly improve through 2027	Low; management itself says “slow and steady”

The tenfold conversion deserves a second look. 16 to 160 fibers is the theoretical ceiling for a 100% optical scale-up; actual penetration is what matters. Weeks knows the market cares about timing, so he spoke with full conviction: “we are on the inside of that ecosystem.” But he gave no penetration assumption for 2028. For which paths scale-up optical interconnect may take and where each hits its bottleneck, see After copper runs out for AI: seven paths for scale-up optical interconnect.

6. Supply Chain and Customer Signals

• Anthropic’s S-1 is cited as demand evidence. Weeks cited the S-1 Anthropic filed last week, both on CNBC and on the call, with three takeaways: growth, the implied profitability of inference tokens, and “compute is the number one constraint.” He positioned Corning as the network that connects that compute. It is the first time a fiber supplier has used an AI model company’s financial filing to argue that demand is durable.

• Customers are mainly hyperscalers, and they are willing to sign long-term deals. Corning signed four more multi-year, multibillion-dollar agreements this quarter (including Verizon and AT&T). Previously disclosed long-term agreement partners include Meta, Amazon and NVIDIA.

• Chinese fiber is not a threat for now. A Wolfe analyst asked about Chinese fiber. Weeks said more and more fiber is integrated into complex connectivity assemblies, so shipping and tariff costs rise with integration, making it far harder to ship than bare reels of fiber; he sees the share of Corning’s markets exposed to Chinese fiber actually falling. But he added that Corning is “always preparing for the moment.”

• LEO satellites are a plus, not a threat. Asked about Starlink moving into terrestrial networks, Weeks said satellite data ultimately has to land in fiber networks, and he is broadly positive.

• The Apple partnership expands. From the end of 2026, all cover glass for iPhone (including iPhone Duo) and Apple Watch sold worldwide will be made at the Harrodsburg, Kentucky plant.

At its 2026 investor day, Marvell likewise named scale-up optics as the one variable starting from zero. The chip side and the fiber side are describing the same timeline, which is a cross-check, and also means everyone is placing the same bet.

7. Other Businesses and Risks

• Glass Innovations is weighed down by memory prices. Sales fell 3% and net income fell 10% YoY. In Q3, TV units were down 4% and notebooks and handhelds about 20%, as rising memory prices squeezed device shipments. A display glass price increase has been announced; it starts in Q4 with little impact, mostly showing up in 2027. The size was not disclosed; management only said it is “similar to what we’ve done the last few times.”

• Yen and hedge ratio. The Q3 yen hedge ratio was 94%, down from 98% last quarter. Management said there is no plan to lower it quarter by quarter, and the long-term plan already stress-tests the yen at ¥150.

• Solar is still losing money. Sales grew 37%, but the segment lost $7M in the quarter as wafer and module lines ramp slower than expected. Section 232 tariffs improve pricing for U.S.-made solar products, but profits depend on yields and utilization improving. Solar remains the biggest drag on company-wide margins.

• ATM equity program. A few months ago Corning registered a $2B at-the-market equity program; the CFO called it “housekeeping” with no intention to use it now.

8. Summary

Corning proved two things this quarter: fiber demand growth no longer depends on GPU count but on fiber density per GPU; and customers willing to prepay to lock in capacity show that supply, not demand, is the bottleneck.

Three metrics to track next quarter:

1. Optical segment net margin: whether 22.4% holds or rises, the most direct evidence of new-product pricing power.

2. Net customer deposits in the cash flow statement: already a net outflow in Q3; if it widens, check whether free cash flow is really “smoothed” as management says.

3. A new DCI target and scale-up ecosystem signals: management promised a new DCI number later; for scale-up, watch whether real optical design wins land in the second half of 2027.

Once customers start paying you to build the plant, you are not selling fiber anymore. You are selling guaranteed capacity.

This article is for technology and industry trend analysis only and does not constitute investment advice.