Corning Core EPS Up 36%, Yet the Stock Sits Below Two Moving Averages

The headline results:

String this year's three quarters together and the curve bends upward. In Q1, core sales were $4.35 billion, core EPS $0.70 and core operating margin 20.2% (company 8-K, 2026-04-28). In Q2, core sales were $4.74 billion, core EPS $0.78, core gross margin 39.6% and core operating margin 20.9% (company earnings release, 2026-07-28). By Q3: core sales of $5.08 billion, EPS of $0.91 and an operating margin of 22.7%. Revenue growth went from 18% to 17% to 19% — broadly steady — but EPS growth jumped from 30% and 30% to 36%. Margins stepping up every quarter tell you each new dollar of revenue is far more profitable than the old business. Optical Communications went from $2.07 billion in Q2 to $2.38 billion in Q3, roughly 15% sequential growth, and it supplied the lion's share of the company's 7% sequential gain.

So who is dragging? Glass Innovations (display glass and specialty materials reported together) brought in $1.506 billion, down 3%. Automotive delivered $470 million, up 4%. Life Sciences and Emerging Growth businesses posted $290 million, up 1%, and lost $28 million. Solar revenue grew 37% to $436 million but still lost $7 million; in the company's own words, the capacity ramp has been "slower than expected" (company 8-K, 2026-10-27). Put simply, nearly all of Corning's growth today rests on one leg — optical communications — while the other four segments combined are running in place.

The company doesn't give a full-year number. Instead it offers its longer-range Springboard targets: an annualized revenue run rate of $20 billion by end-2026, $30 billion by end-2028 and $40 billion by end-2030 (company earnings release, 2026-07-28). Use that yardstick to work out the hurdles:

Q4 guidance implies only about 3.3% sequential growth, slightly below the 4.6% average needed. That's acceptable in a seasonally soft fourth quarter, but from 2027 on the company needs to deliver 4%–5% or more sequential growth every quarter. With the other four segments together barely growing, that load falls entirely on optical communications. The single number to watch in the next quarterly report is the sequential growth rate of Optical Communications revenue. Above 7%, the $30 billion target has room to spare; back below 3%, the market will start discounting the Springboard slope.

Demand-side evidence got more concrete this quarter. Management wrote in the 8-K: "In the third quarter, Corning secured four long-term, multibillion-dollar customer agreements, including with Verizon and AT&T. The company's growing number of major customer agreements underpins the growth outlined in Springboard." (company 8-K, 2026-10-27) Add those four agreements to the Meta contract worth up to $6 billion signed earlier this year, similarly sized long-term deals with two other hyperscalers in Q1 (company 8-K, 2026-04-28), and the long-term partnership with NVIDIA announced in May, and Corning's fiber order book now stretches from telecom carriers to nearly every major buyer of AI computing power. The company also says long-term customers are sharing the cost and risk of capacity expansion, with U.S. domestic fiber capacity set to rise by more than 50% and optical connectivity capacity to grow to ten times its current level (company 8-K, 2026-07-28; NVIDIA partnership announcement, 2026-05).

Corning is a fairly capital-heavy business: draw towers, preforms and furnaces all cost real money, and the NVIDIA partnership commits it to three new plants in North Carolina and Texas. Is the expansion funded with money it earns, or money it borrows?

The cash story hangs together. Nine months of capex consumed only a bit over 40% of operating cash flow, and the leftover free cash flow covers the dividend with plenty to spare. Q3 free cash flow of $661 million against $680 million of GAAP net income is close to one-for-one conversion, meaning the growth on the income statement is arriving as actual cash. The reason expansion hasn't crushed cash flow is that long-term customers share the build-out cost — the 8-K says so plainly. The catch is that the form of that cost-sharing (prepayments, capacity reservation fees or pricing terms) isn't broken out, so readers can't verify what it looks like on the balance sheet. Net debt of about $5.6 billion is no strain at Corning's current pace of cash generation.

On dilution, the market tends to overstate the NVIDIA deal. Media reports say NVIDIA will invest about $500 million in Corning and receive warrants (Reuters and WSJ reports, 2026-05). Against a market value of roughly $108.6 billion, even if all of it converted to equity, existing shareholders would be diluted by only about 0.5%. Public reports differ on the warrant exercise terms and no full company-level disclosure could be found, so we don't take the math further.

The gap worth unpacking is the one between GAAP and core:

Don't panic — this gap isn't new. Corning has used constant-currency reporting for many years, and management's incentives and guidance are both tied to core numbers. The issue is that the gap has long run in one direction, with core above GAAP. If the yen keeps weakening, repricing as hedges roll off would leave GAAP numbers lagging further behind, and the 58x P/E the market calculates on GAAP would diverge sharply from the multiple on core earnings. That point feeds directly into the valuation in Chapter 5.

Corning goes deep in the fiber chain. From preforms (the high-purity glass rods fiber is drawn from) through drawing, cabling and the connectivity hardware inside data centers, most steps happen in its own plants. That concentrates its upstream dependence on a handful of raw materials and inputs:

The roster of downstream anchor customers has grown fast over the past ten months: Meta's multi-year supply agreement (up to $6 billion, 2026-01), similarly sized long-term deals with two other hyperscalers in Q1, the long-term NVIDIA partnership (2026-05), and four multibillion-dollar long-term agreements in Q3 including Verizon and AT&T. The company doesn't disclose revenue share by customer — not available — but with optical communications at roughly 47% of total revenue, the top cloud companies now carry more weight with Corning than any traditional telecom customer.

How much of these contracts is firm? The one most worth unpacking is Meta's "up to $6 billion." That's a ceiling — the most Corning can collect over the life of the contract — and the committed portion isn't public. The four Q3 agreements are described only as "multibillion-dollar," with no individual amounts and no disclosure of minimum purchase volumes or take-or-pay terms. Part of the contracts' firmness shows in customers' willingness to share expansion costs: when a customer pays for capacity, it's worried about not being able to buy enough. But the cost-sharing terms carry no detail either, so readers should treat this as a directional signal only.

Where is the weakest link? In the near term, it's solar. The new chain running from Hemlock polysilicon to wafers has ramped below expectations for two straight quarters and was still losing money in Q3. Management is counting on Section 232 tariffs to improve pricing and demand for U.S.-made solar products; if this leg drags on longer, it will keep eating into core EPS growth. Longer term, the risk is technology. Once co-packaged optics (CPO, which places the optical engine right next to the switch chip) goes mainstream, how racks are wired inside will change, and the industry still disagrees on whether fiber usage goes up or down. Corning is betting that CPO also needs large amounts of high-density fiber connectivity — a call that won't be tested until mass deployment in 2027 to 2028.

Among electronic-component companies, the ones genuinely sharing Corning's AI-connectivity meal are Amphenol, Celestica, Fabrinet, TTM and TE Connectivity. Their Q3 readings (compiled from each company's 2026Q3 disclosures):

Side by side, Corning's overall growth sits in the bottom half of this group of AI-connectivity beneficiaries. The reason was covered above: four other segments with zero combined growth dilute optical's 44% down to 18%. Take Optical Communications on its own and it's nearly level with Fabrinet's 45%, slightly behind Amphenol, and well behind assemblers like Celestica. The difference is profit quality. Corning's core operating margin is 22.7%; Celestica's is just 8.5%. The former sells materials protected by patents and process know-how; the latter earns scale money from assembly. Directionally the whole group agrees: connectivity demand from AI data centers is still accelerating.

How does the market price report cards like this? Among peers that have already reported this quarter, the one whose earnings reactions have tracked Corning's most closely is Knowles. It reported after the close on October 22, beating EPS estimates by 6.7% on surging demand and new products, and gave full-year-basis fourth-quarter guidance. The stock still fell 3.6% the next day, against an options-implied move of ±10.0% beforehand (peer earnings-reaction data, 2026-10-27). The drop used up only a bit more than a third of the implied move: the market's response to a "beat but not dazzling" report was restrained — no panic, no reward. Knowles and Corning moved in the same direction after earnings 6 times over the past 8 quarters, with a correlation of 0.68, but this quarter there is only that one peer sample, so it's of limited use for predicting Corning's reaction. All it really says is that the market currently isn't rewarding beats in component stocks.

Corning's own options priced an implied move of ±7.1% for this report. Over the past 8 quarters, the actual move exceeded the implied move only 2 times, with a median implied move of ±6.0% and a median actual move of 4.8% (peer earnings-reaction data, 2026-10-27). In other words, Corning is usually "all thunder, little rain" on earnings day, and option sellers have long had the edge. With guidance 5.7% above consensus this quarter, breaking that habit would require the market to actually trade the expectations gap over the following days.

The GAAP trailing P/E of 58x no longer works. Its denominator includes low-base quarters from early 2025 and is depressed by currency hedges and restructuring charges, so it systematically overstates Corning's multiple. A PEG of 0.2 is equally distorted, because 256% earnings growth reflects a rebound off a low base and can't continue. The right yardsticks are forward P/E on core EPS, alongside enterprise value relative to annualized core sales:

What does 37.5x mean? In late August, the market was paying close to 88x trailing GAAP earnings for Corning. That's now down to 58x, which works out to under 40x on a forward core basis — the pullback squeezed out a good chunk of the froth. If 2027 core EPS grows 25% from $3.37 (Springboard promises profits growing faster than its 19% revenue CAGR; this is a scenario assumption, not company guidance), that's about $4.2, and the current price would equal roughly 30x 2027 earnings.

That multiple means the market has already prepaid a full year of 2027 growth and is assuming Springboard gets at least halfway to $30 billion. As long as optical keeps delivering high-single-digit sequential growth and margins keep stepping up, earnings can gradually grow into 30x. If sequential growth drops below 3% and EPS growth falls back to around 20%, the market would likely push the forward P/E back toward 25x, leaving no cushion at today's price. The valuation isn't cheap enough to buy with your eyes closed, but it has moved from "pricing in two or three years ahead" back to "prepaying one year."

Fiber is a classic capacity-cycle product. Preforms and draw towers take 18 to 24 months from approval to full output, so when demand shifts suddenly, supply can lag for a long time. The last cycle peaked in 2018; Chinese centralized-procurement prices then halved and the industry spent nearly five years cutting capacity. That's the backdrop for the sudden price surge in 2026.

Three kinds of readings help pin down the position:

Our call: mid-upswing. Prices have already risen several-fold, so the early stage is behind us. But new capacity won't come online until 2027 to 2028, and TrendForce's view that tightness lasts through 2028 matches Corning's expansion timeline, leaving at least a year and a half before oversupply. Picture a restaurant with ten tables and a line suddenly out the door: the owner is rushing to add a second floor, but until the renovation is done, prices and table turnover can only go up.

Two trackable early signals of a peak:

This chapter covers weekly signals that govern timing over the next few weeks. Readers planning to hold for a year or more can skip it.

The fiber-optic communications sub-sector was dropped into the weakest money-flow tier on October 19 and had stayed there 7 straight days as of October 27. It sits in the Lagging quadrant of the relative-strength chart with a money-flow reading of -20, a mild outflow; its parent sector, Communication Services, is also in the weakest tier (sector money-flow data, 2026-10-27). The sub-sector has seen net outflows for 4 straight days. Net volume flow estimated from price and volume totals -127.7 million shares over the past 5 days, yet +1.4 billion shares over the past 20 days, while the sector's 5-day price change is just +1.0% (sector money-flow data, as of 2026-10-27). In trader-speak: a lot of stock changed hands on the way in over the past month, and in the last week some holders have started selling into bounces. Price didn't run ahead of the money; it stood still as the money left.

Where did the money go? As of 2026-10-27, the sub-sectors with the strongest inflows were beauty and personal care, electric utilities and IT services; the heaviest outflows were in lithium batteries, semiconductor deposition and etch, and bitcoin miners. The broad rotation signal leans risk-off, and the top three rotation pairs all show money moving out of tech (XLK) into materials, consumer staples and real estate (market rotation data, 2026-10-27). The rotation clusters now forming are industrials, software and IT services within tech, and travel and cruise lines within consumer discretionary. Neither Communication Services nor the fiber sub-sector belongs to any of them — in other words, no meaningful wave of rotation money is looking for fiber right now.

The macro backdrop is tight too: net liquidity at $5.80 trillion, down 1.2% week over week; the 10-year Treasury yield at 5%; SPY at 769.71, roughly flat with its 20-day average; and an overall market state of neutral-to-weak, with a reading of -16 (macro liquidity and market-state data, 2026-10-27). In this kind of environment, richly valued growth stocks are the first to be swapped for defensives.

Historical statistics offer a less gloomy floor: in the 20 trading days after a comparable sub-sector falls into the weakest tier, the median return is 0.0%, with a 55% probability of a gain (3074 samples, 2021-04 to 2026-09). The weakest tier on its own doesn't predict further declines. It's more of a status label — "the money isn't here for now" — that tells you whether to hurry in the short run, and says nothing about the long-run direction.

Long term (quarters to years): worth following. Corning's optical business is in the middle of a capacity cycle driven by AI data centers, with supply expected to stay tight at least through 2028. The order book keeps thickening, from Meta and NVIDIA to Verizon and AT&T. Core EPS growth has topped 30% for three straight quarters and is still accelerating, profits and cash are arriving together, and expansion is funded mainly by internal cash and customer cost-sharing. A forward P/E of 37.5x on 2026 core EPS — about 32x on annualized Q4 — implies a PEG of about 1.0, back from August's extreme premium to a range earnings can grow into. Three trade-offs come with it: the other four segments combined aren't growing, so the whole company stands on optical alone; the roughly 14% EPS gap between core and GAAP can magnify valuation swings when sentiment sours; and big contracts like Meta's are ceiling figures whose committed portion can't be verified. The way to hold it is to follow the checkpoints. In each quarterly report, check whether optical sequential growth stays around 7%, whether core operating margin keeps stepping up, and whether new long-term contracts keep coming. If all three hold, keep holding; if sequential growth drops below 3%, reassess.

Short term (days to weeks): don't chase — wait for the money to return. The stock is about 16% below its 50-day moving average and about 18% below its 200-day, so the intermediate trend has turned down. Its sub-sector has been in the weakest money-flow tier for 7 straight days with 4 consecutive days of net outflows, broad rotation leans risk-off, money is moving from tech into materials and defensives, and fiber isn't part of any rotation cluster now forming. Guidance 5.7% above consensus can put a floor under the stock, but with money leaving, earnings alone will take time to push it back above its moving averages. What short-term traders should wait for is the sub-sector's money-flow reading turning positive and leaving the weakest tier, and the stock reclaiming a key moving average — at least one of the two before acting.

The second half turns these two verdicts into conditions you can check daily, with three sourced reference prices.