The obvious question about a supply agreement is what it is worth a year. It is the question the announcement will not answer, and going after it anyway is how a specific dollar figure ended up attached to this deal within hours of it being published — a figure that is real, and is nine years old.
The question that can be answered is narrower and more useful: has anyone ever published enough about a Corning fibre contract to price one? The answer turns out to be yes, once, by these same two companies in 2017 — and working out what that disclosure does and does not support is what tells you both where the circulating number came from and why it cannot be right.
What Verizon and Corning actually published
On 8 September 2026 the two companies announced a supply agreement for "80+ million miles of high-density optical fiber and connectivity solutions from 2027 to 2032." The release calls it a "multi-billion dollar agreement." It names Corning's Contour Flow Cable, ties the volume to Verizon's target of 40–50 million broadband passings and to the long-haul corridors behind its AI Connect strategy, and quotes Kyle Malady, CEO of Verizon Business, saying the volume lets Verizon "continue building the network of the future at an unprecedented scale."
No dollar figure appears anywhere in it. Reuters, reporting the same day, put it plainly: the financial terms were not disclosed.
So the public record contains a quantity — 80 million miles — a six-year term, and an adjective.
The $1.05 billion is the headline of a press release from April 2017
Within hours a specific figure was circulating: a $1.05 billion three-year minimum purchase agreement, which divides tidily into $350 million a year.
That is not an estimate anyone made. It is, word for word, the 2017 deal. Corning's release of 18 April 2017 was headlined "Verizon Agrees to $1.05 Billion Three-Year Minimum Purchase Agreement with Corning for Next-Generation Optical Solutions," and its terms were: "Corning to provide and Verizon to purchase up to 20 million kilometers (12.4 million miles) of optical fiber each year from 2018 through 2020, with a minimum purchase commitment of $1.05 billion."
Same two companies, same product, different decade. The term in the current agreement is six years, not three, and the dollar amount does not exist.
The 2017 contract is the only one that published both numbers — and even it does not give a price
This is where the older release earns its keep, because it is the single point in the public record where a Corning fibre agreement carries a quantity and a dollar amount.
It still does not produce a price, and the reason is worth being precise about. The $1.05 billion is a minimum purchase commitment. The 12.4 million miles a year is a maximum — "up to." A floor divided by a ceiling is not an average; it is a bound. Over the three years the agreement covered, the maximum quantity was 37.2 million miles, so the realised average price cannot have been below:
$1.05bn ÷ 37.2m miles = $28.23 per fibre mile.
That is a hard floor on what Verizon paid in 2018–2020, and it is the only price per mile these two companies have ever jointly implied. It is not what they paid — they may well have paid more, since the dollars were a minimum and the miles a maximum.
Which is what breaks the recycled figure
Put $1.05 billion against the new agreement's 80 million miles and the implied price is $13.12 a mile — 46.5% of the 2017 floor.
For that to be right, Corning would have to be selling Verizon fibre at less than half the lowest price its own 2017 contract mathematically permits, on a product the release describes as high-density, with connectivity solutions bundled in. The figure does not fail because it is unsourced. It fails on arithmetic against the companies' own history.
Run it the other way and the 2017 floor puts 80 million miles at about $2.26 billion, or roughly $376 million a year over six years — consistent with the "multi-billion dollar" the release does use, and about 5.2% of the segment's $7.271 billion of trailing-twelve-month revenue through the June quarter.
Treat that $2.26 billion as a benchmark in 2017 pricing rather than a floor on the new deal. Fibre pricing moves, generally downward with scale, and six years of it sits between the two contracts. What survives the caveat is the comparison: a figure implying half the 2017 floor is going the wrong way.
One more thing the two releases put side by side. The 2017 agreement capped Verizon at 12.4 million fibre miles a year. The new one commits to at least 80 million over six years — 13.3 million a year, about 7.5% above the old ceiling. Six years at slightly more than the 2018–2020 maximum annual rate is a substantial contract. "Unprecedented scale" is doing some work in that quote.
Four named agreements, four different units
Corning's optical thesis now rests on four named multi-year agreements, and the reason nobody can add them up is that no two are quoted the same way:
- Meta — "multiyear, up to $6 billion." A dollar ceiling, with no quantity and no annual profile.
- Amazon — "multiyear, multibillion-dollar." An adjective.
- NVIDIA — a capacity commitment: Corning expands US optical connectivity capacity 10x and US fibre capacity by more than 50% across three new plants in North Carolina and Texas. No dollars, no base to apply the multiples to.
- Verizon — 80+ million miles over 2027–2032. A length and a term.
A ceiling, an adjective, a multiple and a length. There is no common denominator, so the forward book cannot be summed — not by us, and not by anyone reading the same four releases. That vacuum is what a nine-year-old number rushed in to fill, and it is the same failure this site has documented before, where a circulating ranking divided quantities that were never in the same units.
The split matters too. Corning names two streams inside the segment — Carrier Networks and Enterprise Networks — and Verizon is the first of the four agreements to land in Carrier; Meta, Amazon and NVIDIA all sit in Enterprise. Corning no longer publishes a dollar split between the two, only growth rates, with Enterprise up 65% year over year in the June quarter. So the one structural fact that would let you allocate the four agreements is also unpublished.
Two-thirds of our valuation sits on the segment that discloses least
The disclosure gap matters because of how much rests on it.
Optical Communications was 43.7% of Corning's core sales in the June quarter — $2.072 billion of $4.738 billion — and grew 32.3% year over year while the other four segments grew 7.5% combined. Segment EBITDA margin, derived from the quarterly segment table rather than disclosed, went from 24.78% to 30.98% in a year. It is the same quarter that took Corning above 40 on the Rule of 40 for the first time in fourteen stored quarters.
In our forward model the concentration is starker. The published base is $152.47. Remove the optical vertical entirely and it falls to $49.05 — the segment carries $103.42, or 67.8% of the whole valuation. Two-thirds of the number rests on a segment whose newest and largest-volume agreement discloses no money at all.
And the model does not move, which is a decision already recorded in it. Our optical vertical runs an 8.0% quarterly growth rate decaying toward a 4.5% terminal, with no base shift — no step change for Meta, Amazon or NVIDIA, and none now for Verizon. An agreement with no disclosed dollars and no annual profile cannot produce one without inventing both.
That is not the model being slow: a capacity driver was considered for the NVIDIA expansion and rejected, because "10x" and "more than 50%" come with no base capacity, no completion dates, no utilisation and no revenue per unit — three inputs that would have had to be manufactured. Verizon arrives with the same shape of problem and gets the same answer.
The one place the new agreement does press on the model is the horizon. The 4.5% terminal quarterly rate is set to roughly 19% a year, the sales CAGR Corning states for Q4 2026 to Q4 2030. The Verizon term runs through 2032, two years past the end of that window, and the model holds the rate flat across the gap rather than extending Corning's own guidance into it.
A $238 target arrived the same day, on the same absence of numbers
China Renaissance initiated coverage on 8 September at Buy with a $238 target, and the note body is not public, so there is no segment split or optical valuation to check it against — only the rating and the number.
For scale: the analyst consensus we store, as of 27 August 2026, is an average target of $191.40 across sixteen analysts polled by S&P Global, in a range of $129.00 to $226.00, with the highest individual action we list being Susquehanna's $220 on 29 July. The $238 is 24% above that average and above the top of the stored range. Corning's price snapshot on 27 August was $152.76; the stock rose roughly 6–7% on 8 September on the announcement and the initiation, per same-day press reports.
Worth noting what the stored actions did before this week: five of the eight most recent cut their target while keeping the rating, and Barclays' $129 sits below where the stock was in late August.
What to watch
- Corning's next quarterly filing, for a remaining-performance-obligation figure. A six-year minimum-volume supply agreement can reach RPO disclosure even when the press release names nothing. That single line would convert 80 million miles into dollars.
- Whether an 8-K quantifies the agreement. Neither company has filed one naming a value; a material definitive agreement filing would carry terms the release omitted.
- Whether this is incremental volume or re-papered volume. Verizon has bought Corning fibre continuously since at least 2017. A minimum-purchase agreement can replace existing volume as easily as add to it, and nothing published says which.
- Any Corning figure for current annual fibre output in miles. The model carries "+50% US fibre capacity" as a percentage with no base. Publish the base and 13.3 million miles a year becomes testable against it.
- The Carrier–Enterprise dollar split. Corning gives growth rates and no dollars. Restoring the split is the one disclosure that would let anyone assign the four agreements to the two streams.
The 80+ million miles, the 2027–2032 term, the "multi-billion dollar" wording, the 40–50 million passings and the Contour Flow Cable are from the Verizon and Corning releases of 8 September 2026; Reuters reported the same day that terms were not disclosed. The $1.05 billion minimum, the 20 million kilometres / 12.4 million miles a year and the 2018–2020 term are from Corning's release of 18 April 2017. The per-mile floor, the $2.26 billion benchmark, the $13.12 implied price, the annual-rate comparison and the percentage-of-revenue conversions are our arithmetic on those two releases. Segment revenue, the 43.7% share, the growth rates and the 30.98% EBITDA margin are derived from Corning's quarterly segment table, not separately disclosed. The base valuation, the drop-optical sensitivity and the growth drivers are ours and are assumptions, not forecasts by Corning. The consensus average, range and analyst actions are as polled by S&P Global as of 27 August 2026; the $152.76 price is that date's snapshot. China Renaissance's Buy and $238 target are press-reported; the note itself is not public.
