Nokia

## Notes (comps_pe — hybrid blend, top-down EPS anchored) ### Headline - Implied: **USD 7.74 / 10.50 / 21.58** (low / mid / high) - Reference: USD **7.20** → mid upside **+45.83%** - Blended Fwd P/E CY26E: **21.08x** (80% telco median 15.53x + 20% AI-infra median 43.31x) - NOK CY26E EPS: **$0.498** (top-down; consensus market-implied $0.48 at 15.0x; +3.81% divergence) ### Peer cohorts (Q1 2026 spot) | Cohort | Peer | Spot USD | Fwd P/E | Freshness | |---|---|---|---|---| | Telco-eq | ERIC | ~7.10 | 13.63x | stale (sourced from public IR) | | Telco-eq | CSCO | ~62.50 | 16.64x | stale (sourced from public IR) | | Telco-eq | HPE | 37.58 | 15.53x | fresh 2026-05-22 | | AI-infra | COHR | 377.57 | 41.49x | fresh 2026-05-22 | | AI-infra | MRVL | 196.33 | 45.13x | fresh 2026-05-22 | ### Why the asymmetric range - Low $7.74 = telco-only multiple (15.53x × $0.498) - High $21.58 = AI-infra-only multiple (43.31x × $0.498) - Mid $10.50 = 80/20 blended The width directly reflects the central investment debate: does Nokia earn an AI-infra re-rating from the NVDA stake, or does it stay anchored to telco-eq multiples? Today's market price ($7.20) implies "stays telco" + small discount; mid case implies partial re-rating. ### Why blend 80/20 (not 50/50) Nokia FY25 revenue is ~85% telco-eq. Mgmt's raised 27% AI+cloud CAGR through FY28 takes the AI-mix to ~25%. 80/20 is the forward midpoint between today's mix and FY28 mix. A 60/40 weighting would push mid implied to ~$13.70 (+90% upside). ### Anchor — thin Single BOM row: `rubin/nvl72 / AI-RAN baseband (downstream platform)`, which is a downstream-application annotation, NOT a per-rack physical BOM item. Nokia AI-RAN revenue does NOT scale with raw Rubin GPU shipments — it scales with telco adoption of NVIDIA Arc Aerial RAN Computer, which is a separate go-to-market motion. Confidence kept at `medium` not `high` because of this thin anchor. ### Caveats - NOK spot $7.20 is sub-agent estimate (NOK not yet in weilab `stock_quotes`); run `pnpm stocks:refresh` post-this to populate and confirm. - ERIC / CSCO Fwd P/E are sub-agent estimates from public IR (not from weilab fresh quotes); fresh peer prices would tighten the band. - AI-RAN commercial pricing model unproven — T-Mobile / BT / Vodafone in functional-test phase, not commercial deployment yet. The +49% Q1 2026 AI+cloud growth is mostly data-center networking and Azure-related, NOT AI-RAN yet. - NVDA can sell its 2.90% stake at any time; "NVDA-backed" is brand signal not contractual commitment.