🔆

GLW

💡
Thesis. Corning owns the fibre layer, with Amazon and NVIDIA as anchor partners funding a tenfold US capacity expansion, and the best financial quality in the category — 39.6% gross margin and $1.42B of free cash flow. The trade-off is that you buy display glass and solar alongside the AI story.
Reviewed 11 September 2026, after Q2 2026 results. Next review after Q3 earnings on 28 October 2026.

What it does

Corning makes optical fibre and cable — the glass strands that actually carry the light — plus the connectivity hardware that joins them together. It also makes display glass for TVs and monitors, solar wafers, automotive components and life sciences products.
Optical Communications is now the growth engine, at around 44% of core sales. CEO is Wendell Weeks.
The structural argument is the interesting one. The scale-up network inside a rack is moving from 100% copper to optical. Clusters above 130,000 GPUs need a third optical layer that didn't exist before, and optical content per GPU is expected to rise 1.3–1.5x by 2028 — which means Corning's enterprise business can grow faster than GPU units grow.

Bull case

  • Q2 2026, reported 28 July: Optical Communications $2.07B, up 32%, with segment net income up 77% to $438M
  • Enterprise Networks up 65%, with generative AI product sales growing significantly faster than that
  • Amazon signed a multiyear, multibillion-dollar agreement to supply fibre, cable and connectivity for its US datacentre expansion — the largest disclosed commitment on this page
  • NVIDIA partnership funds the capacity. Corning expands US optical connectivity manufacturing capacity tenfold and US fibre production by more than 50%
  • Best financial quality in the category. Gross margin 39.6% (+120bps), operating margin 20.9% (+190bps), ROIC 14.9%, free cash flow $1.42B
  • The Springboard plan is specific. A $20B annualised run-rate by end-2026, $30B by end-2028, $40B by end-2030, on a 19% sales CAGR — plus a new $10B photonics platform target by 2030
  • Guidance holds up. Q3 core sales $4.9–5.0B (+16%), core EPS $0.85–0.89 (+28%)

Bear case

  • The diversified base mutes the signal. Total core sales grew 17% while optical grew 32%. You are buying display glass, solar and automotive alongside the part you want
  • Some of that base is weak. Solar grew 90% but posted a small loss, and Life Sciences and Emerging Growth both declined
  • Slowest headline growth in the group — 17% against 32–62% for peers. Lower beta cuts both ways
  • The market wasn't convinced. Shares fell on the print despite the beat, because GAAP revenue of $4.51B read as a miss
  • Springboard is a management framework with 2030 targets. Long-dated promises, not contracted revenue
  • Fibre is a bottleneck that commoditises. It is genuinely scarce now, but competitors can add capacity over time in a way they cannot with indium phosphide lasers

Major customers

Optical Communications
  • Amazon — a multiyear, multibillion-dollar agreement announced in 2026 for fibre, cable and connectivity across its US datacentre expansion
  • NVIDIA — a long-term partnership under which Corning expands US optical connectivity capacity tenfold and US fibre production by more than 50%
  • Hyperscalers and datacentre builders generally, plus telecom carriers for the carrier network segment
Display Technologies — LCD glass to panel makers serving Samsung, LG and the TV and monitor supply chain, and ultimately device brands including Apple.
Other — solar wafer customers, automotive OEMs for emissions substrates and Gorilla Glass, and life sciences labs.
🔗
Read-through. Amazon and
🏛️
NVDA are not merely customers here — they are funding Corning's capacity expansion. That is a far stronger commitment than a purchase order, and the same pattern NVIDIA used with the laser makers.
Corning supplies the fibre that
🔆
COHR,
🔆
LITE and
🔆
CIEN systems run over. It sits upstream of most of this category rather than competing within it, which is why its growth is steadier and its beta lower.
The driver worth tracking: the scale-up network inside racks moving from copper to optical. If that transition accelerates, Corning's enterprise segment outgrows GPU units. Display and solar move on entirely unrelated cycles and are the reason the consolidated number understates the AI exposure.

What would change the view

  • Optical Communications as a share of core sales — the mix-shift metric that decides how this name gets valued
  • Amazon and NVIDIA agreements converting into disclosed revenue rather than announced intent
  • Execution on the capacity expansion — tenfold connectivity and +50% fibre, on schedule
  • Progress toward the $20B annualised run rate by end-2026, the first real Springboard checkpoint
  • Evidence of the copper-to-optical scale-up transition at hyperscalers
  • Whether solar turns profitable or keeps dragging

Update log

  • 11 Sep 2026 — Reviewed. No change to the thesis or the verdict.
  • 7 Aug 2026 — Full review after Q2 2026 results. The Amazon and NVIDIA capacity agreements and the copper-to-optical scale-up transition became the centre of the thesis.
Research and education only — not investment advice.