AI & The Global Fiber + Connectivity Expansion — Strategy Framework

AI & The Global Fiber + Connectivity Expansion — Strategy Framework

Telcotank Strategic Research | February 2026 | Strictly Confidential — For Institutional Use Only

A 60-page institutional-grade research framework mapping how AI is creating a once-in-a-decade connectivity infrastructure supercycle — analysing the AI bandwidth boom across fiber, subsea cables, edge data centers, and emerging market connectivity, with a structured investment thesis for infrastructure fund managers, equity investors, and sovereign development finance allocators.

AI is not just a compute story — it is a bandwidth and physical infrastructure story. This framework quantifies the $602B hyperscaler CapEx surge by 2026, maps the structural shift from legacy north-south networking to east-west AI GPU fabric traffic, and identifies the highest-conviction investment opportunities across the full connectivity infrastructure value chain.

Executive Summary: Four Structural Tailwinds

• Supply-Demand Inversion: AI clusters require 48x the bandwidth of traditional cloud workloads. DC bandwidth purchasing surged 330% in 2024. Existing fiber networks cannot absorb this demand without significant new investment — dark fiber and IRU contracts are being signed 20 years forward.
• Hyperscaler Disintermediation of Telcos: Google (18 subsea cables), Meta ($10B global cable), and Microsoft (MAREA) are bypassing legacy telco infrastructure — bifurcating the market and creating valuation opportunity in neutral fiber carriers (Zayo, Lumen).
• Geopolitical Redundancy Premium: Red Sea cable cuts (Feb 2024) disrupted 25% of Asia-Europe traffic. Taiwan Strait incidents increased 67% in 2025. Cable route diversification is now a national security imperative creating regulatory tailwinds for new infrastructure investment.
• Emerging Market Convergence: Sub-Saharan Africa (Equiano 144Tbps, 2Africa 180Tbps), Southeast Asia, and Latin America are undergoing simultaneous connectivity upgrades via subsea cable, LEO satellite (Starlink 72% market share), and terrestrial fiber backed by $71.7B IFC and $48.9B DFC development finance.

What This Framework Covers

I. AI Traffic Growth & The Bandwidth Imperative

Deep analysis of how AI is rewriting bandwidth economics. AI training clusters generate massive east-west GPU-to-GPU traffic — a single rack with 16 H100 GPUs produces 400 Gbps of east-west traffic versus near-zero for equivalent traditional cloud workloads. Key data points: $602B aggregate hyperscaler CapEx projected 2026 (Amazon $160B, Google $120B, Microsoft $110B, Meta $90B); 75% AI-focused; DC network bandwidth purchasing up 330% in 2024; 53% of DC operators expect AI traffic dominance within 2–3 years; the 400G → 800G → 1.6T upgrade cycle spanning 2022–2028 across GPU generations V100 through B200/GB200.

II. Hyperscaler Private Fiber Builds

Comprehensive mapping of the $10B+ annual private fiber infrastructure race. Google (18 subsea cable investments including Curie, Dunant, Equiano, Firmina, Grace Hopper); Meta (2Africa 180Tbps live Nov 2025, new $10B 40,000km around-the-world sole-owner cable — first hyperscaler-owned global cable); Microsoft (MAREA 200Tbps, NCP consortium); Amazon AWS (100 Direct Connect locations, 400G-ready). Dark fiber market analysis: $7.45B (2024) → $15.19B (2030) at 13–14% CAGR. Zayo Group (90,000 route miles, $1B AI contracts 2024, $3B pipeline post-Crown Castle $4.25B acquisition); Lumen Technologies (400,000 route miles, $150B replication cost, 20-year IRU contracts). Optical networking DWDM supercycle: $5.2B (2024) → $24.7B (2035) with Ciena (50% North America share), Nokia post-Infinera (19% global), and Coherent Corp as primary beneficiaries.

III. Subsea Cable Bottlenecks & Geopolitical Risk

$31.7B market (2024) growing to $44.3B by 2030 at 5.8–6.3% CAGR. Analysis covers the wave of new cable systems (2Africa, MAREA, Equiano, Apricot, Grace Hopper, Blue Raman); the oligopolistic wet-plant manufacturing market (ASN 34%, SubCom 19%, NEC 10%, HMN Technologies 10% — combined 53% concentration); geopolitical cable risk (Red Sea disruptions, Taiwan Strait incidents rising from 3 per year to 5 in 2025); the structural shift from consortium to sole-owner cable models; and the chronically under-supplied repair market (only 70 cable ships globally with 6–8 month repair queues).

IV. Edge Data Center Clustering

$13.8B market (2024) growing to $52.1B by 2030 at 23.6% CAGR. AI inference latency requirements (≤50ms round-trip for interactive AI applications) make edge DC proximity to population centers non-negotiable. Coverage includes: Microsoft Azure Edge Zones (85 locations globally); AWS Wavelength (19 US metro areas, 75% US population coverage); Google Distributed Cloud; colocation giants Equinix (260 IBX locations, 21 xScale facilities, 415MW leased, $15B JV with Morgan Stanley) and Iron Mountain (33% YoY DC revenue growth); and the private 5G enterprise industrial AI stack deployment pattern.

V. Emerging Market Bandwidth Gaps

Africa, Southeast Asia, Latin America, and the Middle East are the decade-long connectivity investment themes. Africa: Equiano (144Tbps, Google, landed Togo March 2024) and 2Africa (180Tbps, 45,000km, 46 landing stations in 33 countries, live Nov 2025). Southeast Asia: Singapore (99.75% FTTH, AI hub), Vietnam (80% FTTH, 5G in 63 provinces), Indonesia (65% FTTH, 17 IXPs). Middle East: MENA fiber market $17.4B (2025) → $22.8B (2030) at 10.3% CAGR; Saudi Arabia Vision 2030 (3.5M new FTTH households, $8.7B public digital contracts, HUMAIN $40B AI fund via PIF); UAE (99.3% FTTH penetration — highest globally; MGX $100B AI infrastructure fund). LEO satellite: Starlink (72% market share, Q2 2025) and AST SpaceMobile (50 MNO partnerships, 2.8B subscriber reach, D2C US launch 2026) — $2.5B → $43.3B direct-to-device market at 32.7% CAGR 2025–2034.

VI. Investment Thesis & Opportunity Framework

Three-tier risk-return framework across the fiber value chain:

• Tier 1 — Core Infrastructure (8–12% IRR, 10–20 year horizon): Dark fiber IRUs (Zayo, Lumen 20-year contracts); neutral carrier fiber post-Crown Castle. Hyperscaler AI demand locked in via 20-year IRU contracts creates bond-like cash flows with AI growth optionality.
• Tier 2 — Growth Infrastructure (15–25% IRR, 5–10 year horizon): Colocation edge DCs (Equinix xScale, Iron Mountain); subsea cable vendors (ASN/SubCom); optical networking (Ciena, Coherent, Nokia post-Infinera); telecom infrastructure carve-outs (Vantage Towers model). The 400G–1.6T upgrade cycle and geopolitical redundancy premium drive structural demand.
• Tier 3 — Emerging Opportunities (25–40% IRR, 3–7 year horizon): LEO satellite (AST SpaceMobile, Eutelsat OneWeb); Africa/SEA neutral fiber operators; AI-native connectivity startups. Development finance (IFC $71.7B, DFC $48.9B) provides risk scaffolding.

VII. Risk Framework & Scenario Analysis

Four risk dimensions mapped across probability and impact: Geopolitical (Taiwan Strait escalation, Red Sea continuation, US-China tech decoupling); Technology Substitution (LEO satellite vs terrestrial fiber, AI model efficiency reducing bandwidth demand, FWA/5G vs FTTH); Regulatory/Policy (US BEAD delays, cable landing security requirements, AI data residency rules, BIS export controls); and Valuation/Market (infrastructure PE multiple compression, hyperscaler CapEx cycle reversal, dark fiber oversupply risk). Three scenarios modelled: Bull Case (35% probability, Tier 2 IRR 25–35%); Base Case (45% probability, Tier 2 IRR 15–22%); Bear Case (20% probability, Tier 2 IRR 5–14%).

Top 5 High-Conviction Investment Opportunities (2026–2028)

1. Zayo Group Dark Fiber IRUs — 90K route miles post-Crown Castle, $1B AI contracts 2024, $3B pipeline, 13–14% CAGR dark fiber market, 20-year hyperscaler IRU backlog. Target IRR 10–14%.
2. Ciena Optical Networking (CIEN) — 50% North America market share, WaveLogic 6e (800G/1.6T-ready), AI DC optical interconnect market $10B→$30B by 2030. Target IRR 18–25%.
3. Equinix xScale (EQIX) — 260 IBX locations, 33 countries, 21 xScale facilities, 415MW leased, 23.6% CAGR edge DC market, $15B JV with Morgan Stanley. Target IRR 12–18%.
4. SubCom/ASN Subsea Wet Plant — 53% combined wet-plant market share, backlog growing on geopolitical redundancy demand, Meta $10B and Google private cables in queue. Target IRR 15–22%.
5. AST SpaceMobile (ASTS) — 50 MNO partnerships covering 2.8B subscribers, D2C US launch 2026, $2.5B→$43.3B LEO market at 32.7% CAGR, Vodafone JV and AT&T partnership. Target IRR 25–40%.

Framework Specifications

• Publisher: Telcotank Strategic Research
• Date: February 2026
• Length: 60 pages across 7 sections + appendix
• Format: Downloadable PPTX (PowerPoint presentation)
• Coverage: 5 market verticals, 8 investment plays, 12 countries/regions, global
• Data Sources: Mordor Intelligence, Grand View Research, GMinsights, DellOro Group, S&P Global, TeleGeography, hyperscaler earnings calls, GSMA Intelligence
• Market Data: $602B hyperscaler CapEx (2026), $52.1B edge DC (2030), $44.3B subsea cable (2030), $15.19B dark fiber (2030), $43.3B LEO satellite (2034)

Digital asset (PPTX) will be delivered upon purchase. License: Single-entity use. Strictly Confidential — For Institutional Use Only.