Ribbon Communications (RBBN) vs ADTRAN Holdings (ADTN) — Tier-2 Optical Through the AI-DC Interconnect Lens

Buy-side deeper-report · comparative two-name coverage · neutral survey · 2026-06-16

The central question. Are RBBN and ADTN cheap tier-2 optical names with real AI–data-center (AI-DC) interconnect optionality the market is missing — or correctly-discounted sub-scale vendors whose AI-DC exposure is too small, and whose risks too large, to matter?

Terms used in this note (defined on first use): DCI — data-center interconnect, the optical links that carry traffic between (and increasingly within) data centers. ZR / ZR+ — industry standards for coherent pluggable optics (transceivers that plug directly into routers/switches), with 400ZR, 800ZR and ZR+ denoting capacity/reach variants. Coherent pluggable — a coherent optical transceiver in a pluggable form factor that lets a router host the optics directly. Scale-across — interconnecting AI data centers across metro/regional distances so multiple sites act as one training fabric. MOFN — Managed Optical Fiber Network, where a hyperscaler contracts a carrier to build/operate optical capacity on its behalf. DPLTA — Domination and Profit-and-Loss Transfer Agreement, the German corporate instrument governing ADTRAN’s control of its Adtran Networks SE subsidiary and the minority squeeze-out. SOTP — sum-of-the-parts valuation.

1. Investment summary / thesis (answer-first)

The evidence-led answer: both are correctly-discounted sub-scale tier-2 optical/access vendors with real but minority AI-DC optionality — not a missed AI re-rating. ADTN has the stronger claim of the two; neither is thesis-grade AI exposure.

The starting point is scale, and it is unflattering for both. Neither Ribbon nor ADTRAN appears in any optical-systems leader tier: on a trailing-four-quarter basis through Q1 2026, Huawei, Ciena, Nokia (including Infinera) and ZTE held more than 80% of optical transport systems revenue between them, and for the ZR/ZR+ IPoDWDM segment the lead vendors are Marvell and Cisco/Acacia — neither RBBN nor ADTN is named ¹². Each company’s optical franchise runs at roughly $0.33–0.38B annually versus Ciena’s ~$5.6B, i.e. about one-fifteenth the scale ³³⁴. A separate sell-side interconnect tiering (Bernstein, 97pp) names component/DSP leaders and neither of these two ⁵. Three independent reads place both firmly in the tier-2 challenger band.

AI-DC exposure is real but a minority of each name, and is more product-positioning than booked revenue. On the model’s analytic build-up, RBBN carries ~$25.0M of AI-DC-exposed revenue (~3.0% of FY2025 sales) and ADTN ~$123.6M (~11.4%) — ADTN roughly 4.9x RBBN. RBBN discloses no data-center revenue line at all, so its figure is inferred small ⁶; ADTN’s larger number is grounded in a cloud/enterprise mix that the company does disclose (21% of FY2025 revenue), optical growth its management attributes to cloud data-center capacity, and an explicit subsidiary-level “AI data centers” attribution ⁷⁷⁸⁹. Both companies’ AI framing leans on product launches and reference customers rather than disclosed AI revenue ¹⁰¹¹.

The AI re-rating accrued almost entirely to the scale leader. Ciena trades at roughly 11.6x EV/Sales; RBBN at ~1.0x and ADTN at ~1.2x got none of it ^34121212.

On the relative call within a neutral survey: ADTN has the stronger claim — more genuine AI-DC exposure, faster optical growth (+26.4% YoY), and a deleveraging turnaround — but it carries the DPLTA squeeze-out overhang, up to ~€304.4M/$351.7M if all minorities elect, per the Q1 2026 10-Q ⁸⁷¹³. RBBN is cheaper but more levered, with a May-2026 covenant waiver and near-zero demonstrated AI-DC traction (zero named hyperscaler wins in FY2025) ¹⁴⁶. Neither is a pure-play AI-interconnect bet; both are optionality-with-baggage ⁵⁶⁷¹⁴⁸.

2. Variant view — what the Street is (and isn’t) missing

The Street prices both as cyclical sub-scale hardware vendors, not AI beneficiaries — and that is the variant the bull case must exploit. Ciena’s ~165% year-to-date move and ~11.6x multiple sit on a verified $5.57B top line growing ~30.6% YoY ¹⁵⁴. RBBN and ADTN lack a comparable disclosed AI revenue line ⁶, and only ADTN shows a demonstrably cloud/DC-driven optical mix ⁷.

So the honest variant is narrow: the market may be under-crediting ADTN’s demonstrated optical pull, not that it has mispriced a hidden AI franchise at either name. A re-rating toward the leader’s multiple would require disclosed AI-DC revenue that the filings do not yet show. The valuation gap is the opportunity; the exposure reality only partly supports closing it ³⁴⁷⁶.

3. Business & economics — the optical sliver and who buys it

RBBN. Optical is the minority of the company. IP Optical Networks generated $333.1M in FY2025 (~39% of the $844.6M total); the larger Cloud & Edge segment ($511.4M, ~61%) is voice/UC modernization with no data-center linkage ¹⁶¹⁶³. Even a fully successful optical story addresses under 40% of Ribbon. The Apollo platform’s AI-DC capability is real — 800G ZR+ scaling to 1.2T coherent, Scale-Across, a distinctive “Rural DCI” angle targeting 67% of US DCI growth via cross-sell to the rural telco installed base — but it is positioned at Tier-1/Tier-2 service providers, not sold direct to hyperscalers ¹⁷¹⁸. The capability is architecturally credible; the channel is mismatched to where hyperscaler spend lands.

ADTN. Optical is the largest and fastest-growing product family: $380.3M in FY2025, +26.4% YoY, with Q4 optical up 33% YoY that management attributes to cloud providers expanding data-center capacity plus enterprise upgrades ⁸⁷. The other two families are broadband-access-oriented and grew on telco demand: Subscriber Solutions $369.1M (+11.6%) and Access & Aggregation $334.4M (+14.9%) ⁸⁸. The 10-K explicitly frames the business around metro optical transport, DCI and broadband access, and names hyperscalers, neocloud and data-center companies in its customer base ⁸. The optical line is the single best-evidenced AI-DC linkage in the pair.

The exposure cannot be pinned precisely, and the imprecision cuts both ways. ADTN’s enterprise+cloud-provider mix was 21% of revenue in FY2025 (25% in Q4) — but that blends large non-AI enterprise (e.g. banking), so it overstates AI-DC; meanwhile MOFN demand (hyperscalers contracting carriers to build for them) books as carrier revenue, so direct hyperscaler pull is split across lines and understated ⁷⁷. The model’s ~11.4% is therefore the midpoint of a wide range, not a disclosed figure. A second, smaller thread: the Oscilloquartz timing portfolio offers a niche GPS/GNSS-independent timing angle for data centers, but it sits in Access & Aggregation, not Optical, and is unsized ⁸.

For context on how much of each name is even optical:

4. Competitive position / moat — the tier-2 reality and merchant-pluggable disruption

The demand backdrop is genuine and large. Optical transport equipment grew to $16B in 2025 (+10%) and is forecast to surpass $18B in 2026 (+16%), DCI-led, with DCI growing roughly twice the overall market and 800ZR/ZR+ pluggables exceeding $1B of revenue in 2026 ¹⁹²⁰²¹. The opportunity is real; the open question for tier-2 vendors is reach, not size.

Two structural forces work against that reach. First, the architecture itself is shifting under them: cloud operators have made router-hosted coherent pluggables their default DCI architecture, shifting an estimated ~$2.5B out of traditional optical transport in 2024 alone ²²²¹. That fault line both displaces the chassis transport RBBN and ADTN historically sold and forces them to compete as pluggable hosts on someone else’s silicon. Second, the scale leaders are consolidating up into the wave: Nokia closed Infinera in February 2025, creating a >$3.6B optical business ²¹. The scale gap is widening, not narrowing.

The outside-in read confirms the inside-out one: sell-side interconnect coverage tiers the investable set around component and DSP leaders (Coherent, Lumentum, Fabrinet, Corning) and names neither RBBN nor ADTN ⁵. The market’s neglect looks considered, not an oversight.

5. Financials & forecast — growth, margins, earnings quality

RBBN — flat top line, weak earnings quality. FY2025 revenue was $844.6M (+1.3%) ²³ with non-GAAP gross margin compressing to 52.3% (from 55.9%) ²³, a GAAP operating loss of $(3)M ²³, and adjusted EBITDA down to $107M ²³. The headline GAAP net income of $39.6M is tax-benefit-driven, not operating — the reader should not mistake it for earnings power ²³. The balance sheet is workable but thin: total debt of $405.1M (term debt plus finance leases) against $96.4M of cash, leaving the $308.7M net-debt figure that frames the leverage discussion below ²⁴²⁴. The FY2026 guide is modest and back-half weighted: revenue $840–875M (midpoint $857.5M, ~+1.5%), gross margin 52.5–53.5%, adjusted EBITDA $105–120M, with IP Optical guided to ~5% growth driven by India/EMEA telco and BEAD broadband subsidies — explicitly not hyperscaler AI-DC ²³²³⁶. The Q4 IP Optical line actually declined 2% YoY and US Federal revenue roughly halved, framing the caution ⁶⁶.

ADTN — real turnaround on non-GAAP/cash, unproven on GAAP. FY2025 grew 17.5% to $1,083.8M, with the non-GAAP operating margin swinging to +4.8% (from -0.3%) and free cash flow up 58% to $60.5M ⁸⁷⁷. GAAP gross margin also improved to 38.3% (from 35.1%), evidence the mix/cost work is reaching the P&L ⁸. But the company remained GAAP loss-making: operating loss $(15.6)M, net loss $(45.7)M — narrowed sharply from 2024’s impairment-driven losses, yet still red ⁸⁸. ADTN gives only Q1‘26 guidance ($275–295M, non-GAAP operating margin 4–8%) and no full-year 2026 guide; the Q4‘25 exit-rate was strong (revenue +20.1%, a sixth consecutive sequential-growth quarter), though component supply (memory/optics) is a flagged risk ²⁵²⁵. The short book-to-ship and absent FY guide are themselves earnings-quality signals. Both names run heavy R&D (RBBN ~21.8% of revenue, ADTN ~18.4%) and ADTN’s trailing ROE is still negative (~-24%) on the GAAP losses — consistent with the picture above and not separately load-bearing ¹²¹²¹².

6. Valuation — comps and the illustrative SOTP gap

On the comparable set, RBBN (0.97x EV/Sales) and ADTN (1.25x) trade far below access peer Calix (2.26x) and the AI-optics beneficiaries Coherent (10.13x) and Ciena (11.63x) ^1212123. RBBN’s and ADTN’s multiples here are computed on FY2025 revenue (the model basis); on the trailing-twelve-month basis used for the peer figures they screen marginally higher, at 0.99x and 1.21x respectively ¹²¹².

EV/Sales basis: RBBN and ADTN computed on FY2025 revenue (model); Calix, Coherent and Ciena on FMP trailing-twelve-month. The two bases differ only marginally for the tier-2 names (RBBN 0.97x vs 0.99x TTM; ADTN 1.25x vs 1.21x TTM) and do not affect the comparison.

The EBITDA multiples carry less signal than EV/Sales here. RBBN’s (18.19x) and the peers’ optically high readings — Ciena 97.54x, Coherent 57.65x — sit on depressed or fast-growing EBITDA denominators, and for RBBN the GAAP/non-GAAP EBITDA-basis issue noted below applies. A sales-multiple comparison is therefore the cleaner lens for the re-rating question ¹²¹²¹².

An illustrative sum-of-the-parts — valuing the AI-DC sliver at 10x sales (in line with where the scale leaders Coherent and Ciena trade, ~10–12x) and the rest at 1x — implies a +30.4% enterprise-value uplift for RBBN and +62.2% for ADTN ¹²¹². This is indicative of the embedded optionality, not a price target: ADTN’s larger sliver drives the bigger gap.

Leverage colors the equity read. RBBN’s net debt is $308.7M, roughly 2.9x adjusted EBITDA, versus ADTN’s ~$125M, roughly 1.24x ²⁴¹². Note a data conflict worth flagging: FMP reports RBBN net-debt/EBITDA at 7.33x, but that is computed on a GAAP/trailing-EBITDA basis that does not reconcile to the ~2.9x on FY2025 adjusted EBITDA of $107M ²³ and net debt of $308.7M — the model adopts the ~2.9x adjusted-EBITDA basis as primary and treats the 7.33x as the FMP-basis artifact ¹²²⁴. The corrected read still leaves RBBN the more-levered of the two, and its cheaper sales multiple is partly a more-levered balance sheet rather than pure neglect.

7. Scenarios — per name, and the asymmetry

ADTN. Bull: demonstrated cloud/DC optical pull continues (Q4 optical +33%) and the EMEA Huawei-replacement opportunity (~$850–860M of annual EMEA Huawei spend in ADTN’s target products, on management’s estimate) converts to share ⁷⁷. Bear: the DPLTA exit cash drain crystallizes while a GAAP-loss relapse returns ¹³⁸. RBBN. Bull: the nascent DCI wins become disclosed revenue. Bear: flat optical, federal softness and covenant pressure compound ⁶⁶¹⁴. The asymmetry favors ADTN on demonstrated traction, but with a larger discrete cash claim (DPLTA) standing against it.

8. Catalysts — next 6–12 months, hard vs soft

Soft (product/reference, no disclosed AI revenue): ADTN’s LiteWave800 — its first product purpose-built for intra-data-center AI infrastructure (800G), with volume production roughly a year out — and a quantum-safe optical narrative; RBBN’s Q1‘26 “several new Data Center Interconnect wins” (no DCI revenue disclosed) and an AWS Strategic Collaboration Agreement ^26101127. The path to a re-rating runs through disclosed AI revenue that does not yet exist.

Hard (dated, testable): ADTN’s Q1‘26 print already landed ($286.1M, +15.5%; optical $237.9M) and DPLTA appraisal proceedings are live; RBBN’s Q1‘26 already showed revenue down 10% to $163M with negative adjusted EBITDA of $(8)M and a May-2026 covenant waiver ^26271314. The next earnings prints, covenant tests and the appraisal outcome are the events that move the names.

9. Risks and what would change my mind — falsifiable trip-wires

For the bull case (would convert optionality to thesis-grade): a named hyperscaler or neocloud optical design win with disclosed AI-DC revenue at either name. RBBN named zero such wins in FY2025; ADTN names hyperscalers in its customer base but discloses no AI-DC revenue line ⁶⁸⁶⁷. A disclosed win or revenue line would flip the call.

Against (would confirm the discount): the router-hosted pluggable shift accelerating faster than DCI wins replace the displaced systems/chassis revenue both depend on. RBBN’s flat-to-declining IP Optical (Q4 -2%) is the early tell ²²⁶²¹.

10. Single-name specifics

RBBN. Market cap ~$489.7M, enterprise value ~$820.4M, share ~$2.78, consensus price target $3.50 ^12121212. The term loan matures in 2029 at SOFR+6.25% — no maturity wall — but a May-2026 covenant waiver (raising max net leverage to 4.50x, stepping to 4.00x) makes leverage headroom the binding single-name risk, alongside high-cost floating-rate debt ²⁸¹⁴.

ADTN. Market cap ~$1,223M, enterprise value ~$1,353.8M, share ~$15.10, consensus price target $18.00 ^12121212. ADTRAN owns just over 70% of Adtran Networks SE; the DPLTA squeeze-out gives minorities a €17.21/share cash put, aggregating up to ~€304.4M/$351.7M if all elect, per the Q1 2026 10-Q (including guaranteed interest, at the 3/31/26 exchange rate) — and German appraisal proceedings could raise it. Funded debt is ~$201.3M of 3.75% convertibles due 2030 plus a $25M revolver ¹³⁸. The DPLTA claim, against an otherwise improving balance sheet, is the binding single-name risk.

The optical exposure lives in a specifically-named subsidiary the reader can track: Adtran Networks SE (ex-ADVA, Munich), which houses the FSP 3000 platform and AOE coherent pluggables, grew FY2025 group revenue 10% to €481.7M and explicitly attributes demand to an “unprecedented investment boom in AI data centers” — the strongest filing-grade AI linkage in the pair, though framed largely via service-provider transport rather than direct hyperscaler sales ⁹⁹.

Implementation/sizing. Both are small-cap and sub-scale; RBBN’s micro-cap size and covenant overhang argue for the smaller position if either is taken, while ADTN’s larger float and cash generation give it the cleaner balance sheet net of the discrete DPLTA claim.

11. Sources

Full provenance — source URLs, document types, publication dates and confidence — lives in the findings sidecar: /Users/johnhendren/Desktop/Projects/vault/raw/research-2026-06-16-rbbn-adtran-dc-networking/findings.jsonl (S001–S087). Derived valuation figures (AI-DC-exposed revenue estimates, the illustrative SOTP uplift, comps) tie to the model /Users/johnhendren/Desktop/Projects/vault/kb/outputs/models/rbbn-adtran-dc-networking-comps_v1_2026-06-16.xlsx.

Key primary sources cited above include: RBBN and ADTN FY2025 results and 10-K filings [S027–S054]; RBBN and ADTN Q1‘26 results ²⁷²⁶; Adtran Networks SE FY2025 annual report ⁹⁹; the ADTRAN Q1 2026 10-Q (current DPLTA Exit Compensation) ¹³; Dell’Oro and Cignal AI optical/DCI market forecasts [S069–S071]; FMP market-data snapshots and stockanalysis.com Ciena verification [S001–S026][S081–S084]; and Bernstein interconnect coverage ⁵.

Sources