The Capex-Revenue Disconnect
Telecom carriers have deployed roughly $200 billion globally in 5G capex over the past seven years, with the largest carriers—Verizon, AT&T, T-Mobile US, Deutsche Telekom, and NTT Docomo—each spending $30–60 billion. Yet the promised revenue uplift has not materialized. Revenue per gigabyte continues to compress. In the US market, Verizon’s wireless service revenue grew less than 3% annually from 2018–2024, despite $120+ billion in capex. AT&T’s postpaid phone ARPU (average revenue per user) has been essentially flat for five years. The math is unforgiving: capex as a percentage of revenue has risen to 15–20% at major carriers, while return on invested capital (ROIC) has compressed. For context, Apple and Nvidia maintain capex/revenue below 5% while delivering 15%+ ROIC. Telecom’s problem is architectural.
The financial trap is simple: deployment of spectrum and network infrastructure costs are sunk, non-recoverable, and front-loaded. Revenue realization depends on demand elasticity that simply does not exist. Consumers and enterprises do not buy incremental data; they buy speed tiers until saturation, then churn to cheaper plans. Operators bet that 5G would unlock enterprise edge computing, automotive, healthcare, and manufacturing use cases. Nearly all of these bets have failed or been indefinitely delayed.
Why 5G Monetization Collapsed
The 5G revenue thesis relied on three pillars: (1) consumer demand for ultra-fast mobile data driving higher ARPU; (2) enterprise adoption of edge computing, IoT, and private networks generating new revenue streams; and (3) spectrum scarcity economics driving pricing power. All three collapsed.
Consumer demand saturation. By 2020, 4G was already delivering 100+ Mbps for streaming, videoconferencing, and web use. Consumers do not experience perceptible benefit from 5G’s theoretical 1Gbps; they cannot distinguish it in real-world use. Carrier marketing claims of “10x faster” were technically accurate but commercially irrelevant. Actual 5G throughput in dense urban areas averages 200–400 Mbps, not 1Gbps. Subscribers saw no reason to pay premium prices for imperceptible speed gains. Average ARPU increases from 5G upgrade remained under $5–10 per month, insufficient to justify $50+ billion in network capex.
Enterprise use-case delays. Carriers projected that automotive (connected vehicles), manufacturing (Industry 4.0), healthcare (remote surgery), and smart cities would drive new revenue. Seven years later:
- Connected vehicles: Adoption remains below 10% of new cars sold. 4G/LTE is sufficient for remote diagnostics and over-the-air updates. Autonomous driving companies use private networks or satellite. No material revenue stream has emerged.
- Private 5G networks: Enterprise customers built them on their own or via neutral-host third parties (AWS, tower companies) rather than buying from incumbent carriers. Carriers expected to be the primary vendor; instead, they became an afterthought.
- Healthcare and remote surgery: Regulatory delays, latency requirements not unique to 5G, and low demand from hospitals have kept this use case negligible.
- Manufacturing: Similar story—enterprises adopted Industrial IoT and edge computing through cloud vendors (AWS Wavelength, Azure Edge Zones), not carrier offerings.
The common thread: enterprises demanded flexibility and control, not carrier-managed services. Carriers were positioned to sell connectivity; enterprises wanted software platforms.
Spectrum scarcity economics evaporated. As of 2025, spectrum supply exceeds demand in most developed markets. The US, EU, and Japan have conducted multiple spectrum auctions; mid-band and high-band capacity is now plentiful. Without scarcity, pricing power disappears. Carriers cannot raise consumer ARPU significantly without losing subscribers to rivals; enterprise customers have alternatives.
The 4G Contrast: Why Monetization Actually Worked
The 4G capex cycle (2009–2016) generated measurable ROI because demand was real and novel. Smartphones were ascendant. App ecosystems emerged: Netflix, Instagram, YouTube, Uber, Snapchat. These services required broadband speeds (10–50 Mbps) that 3G could not deliver. Consumers felt the gap and paid for it. Operators recaptured that investment through ARPU expansion from $30/month (3G era) to $50–70/month (4G era). That 100% lift in ARPU translated to healthy ROI on 4G capex, even accounting for operating cost inflation.
5G launched into a world where data consumption was already saturated and app ecosystems were mature. There was no killer application analogous to Netflix or Snapchat that demanded higher bandwidth. The addressable revenue growth from consumers was near-zero. Carriers poured capital into a network that was technically superior but commercially redundant.
Quantifying the Capex Trap: Carrier-Level Analysis
Verizon: $120+ billion in capex (2015–2024), primarily for 5G deployment. Operating expenses rose from $95 billion (2015) to $135 billion (2024). Wireless service revenue grew from $85 billion to $110 billion—a compound annual growth rate of 2.5%. The incremental revenue of $25 billion over nine years does not justify $120 billion in capex. Even assuming 10% annual capex depreciation over 15 years, the economic cost is $80 billion against $25 billion incremental revenue. ROIC on this capex cohort is well below cost of capital (est. 4–5%, vs. 8–9% WACC).
AT&T: $130+ billion in capex over the same period. Core wireless ARPU remained flat at $60–65. Revenue growth came entirely from M&A (FirstNet contract, Cricket Wireless). Organic postpaid phone revenue was essentially nil. AT&T’s debt rose to $180+ billion; interest expense consumed 20%+ of EBITDA by 2024. Dividend payments required debt issuance, not organic cash flow. The company is in a refinancing trap: it must maintain capex levels to remain competitive (carrier ritual), yet capex generates no return.
Deutsche Telekom: €180 billion cumulative capex (2015–2024) across all regions. German wireline EBITDA margins compressed from 40% to 32%. Capex as a percentage of revenue in Germany rose to 18%. Refinancing costs increased €2 billion annually (2023–2024) due to rising rates and higher leverage. The company suspended share buybacks and held dividend flat, signaling investor concern about capital adequacy.
NTT Docomo: ¥4.5 trillion capex (2015–2024). Japan’s mobile ARPU is among the world’s lowest (¥3,500/month ≈ $25 USD equivalent), yet Docomo maintains capex intensity above 15%. Operating leverage is compressed; Docomo’s net income fell from ¥1.1 trillion (2015) to ¥900 billion (2024) despite flat revenues, entirely due to rising interest and capex depreciation.
Pattern: Every major global carrier exhibits the same dynamic: capex deployed, revenue flat, operating costs rising, ROIC below cost of capital, debt rising, financial flexibility eroding.
The Capex Debt Spiral
Telecom debt markets are pricing in structural stress. Investment-grade telecom credit spreads widened 150–200 basis points from 2020 to 2024, while sovereign spreads tightened. Refinancing risk is material.
AT&T has $40 billion in debt maturing over the next three years. Verizon has $35 billion. Both will refinance in a higher rate environment than they borrowed in (pre-2022). Refinancing cost for a $10 billion tranche rose from 2–3% in 2021 to 4–5.5% in 2024. That 2.5–3% incremental annual interest expense ($250–300 million per $10 billion refinanced) must be funded from operations, since capex already constrains free cash flow.
The 6G horizon deepens the trap. Industry consensus (3GPP, vendors) suggests 6G standardization by 2028 and first deployments by 2030–2032. Carriers are already discussing 6G capex requirements. No executive can credibly say “we will not invest in 6G” without risking spectrum license challenges and competitive obsolescence. Yet 6G will be deployed into the same capex trap: technical superiority, zero incremental consumer demand, and no enterprise killer app.
Refinancing costs will continue to rise if rates stay elevated. If rates fall, capex pressure accelerates (lower cost of capital justifies more capex). Either way, carriers cannot escape.
Operating Cost Headwinds Compress Returns Further
Even assuming capex deployment is optimized, operating expenses are rising faster than revenue.
Energy costs: Network infrastructure consumes 2–3% of developed-world electricity. A typical cell site draws 10–15 kW continuously. Global energy prices are 30–50% higher than 2020 levels (in real terms). Carriers’ aggregate annual energy bill has risen $10–15 billion globally. This is largely fixed; carriers cannot negotiate it away.
Labor costs: Skilled technician and network engineering labor is tight. Wage inflation has pushed average salaries up 15–20% over five years. Carriers employ 50,000–200,000 globally; payroll inflation is significant. Additionally, spectrum auctions and regulatory compliance have added headcount without corresponding revenue lift.
Spectrum licensing: Annual spectrum lease costs to governments have risen as carriers have acquired additional mid- and high-band licenses. In the US, AT&T and Verizon each pay $2–3 billion annually in spectrum license fees. This is a recurring cost floor that cannot be lowered.
Site acquisition and leasing: As carriers densify networks (small cells, distributed antenna systems), real estate costs rise. Rooftop leases, backhaul fiber, and power infrastructure now represent 10%+ of operating expense at major carriers.
The cumulative effect: operating expense growth (3–5% annually) exceeds revenue growth (0–2%). EBITDA margins compress. For example, AT&T’s wireless EBITDA margin fell from 45% (2015) to 38% (2024)—a 700-basis-point compression. That margin erosion consumes the capex return entirely.
The Tech Capex Paradox: Why Apple and Nvidia Are Different
Apple’s capex is 1–2% of revenue; Nvidia’s is 2–3%. Both companies generate 15%+ ROIC. Why the divergence?
Capex accelerates demand. Apple’s capex is primarily manufacturing capacity and supply chain resilience. New capacity is built in response to demand signals (new product launches, market growth). Capex does not precede demand; it follows it. For Nvidia, capex funds AI chip fab partnerships and AI infrastructure, which directly serve booming enterprise AI adoption.
Capex is variable, not fixed. If demand evaporates, Apple can curtail capex with minimal sunk-cost drag. Telecom capex (spectrum licenses, tower builds) is largely irreversible and committed.
Return is measured in months to 2 years, not 10+ years. Apple’s iPhone 16 launch justifies incremental capex in a single cycle. Telecom carriers assume a 10–15 year depreciation schedule for network capex, then pray demand materializes over that horizon. It does not.
Competitive dynamics drive profitability, not capacity utilization. Apple competes on product differentiation and brand; Nvidia on semiconductor performance. Capex supports these. Telecom competes on coverage and price; capex races are zero-sum (every competitor builds equivalent networks) and drive prices downward.
Investor Implications: Valuation Compression and Covenant Risk
Institutional equity investors have already repriced telecom. Valuation multiples (P/E, EV/EBITDA) for telecom are 40–50% below their 2010 levels, even in nominal terms. Telecom trades at 8–10x EBITDA vs. 12–15x for industrials and 20–25x for tech, despite superficially similar credit quality.
Dividend pressure. Dividend sustainability is at risk. AT&T and Verizon both yield 7–8%, well above historical norms. Markets are pricing in dividend cuts. Free cash flow (operating cash flow minus capex) at both companies has compressed to $25–35 billion annually, barely sufficient to cover dividends ($11–15 billion) and debt service ($8–12 billion). Any revenue decline or capex acceleration triggers a dividend cut. Telecom dividend funds have recognized this; selling pressure is evident.
Equity valuation: Telecom is trapped in a low-ROIC equilibrium. As long as capex produces returns below cost of capital, equity holders are destroyed. Telecom equity has underperformed the broad market by 3–5% annually since 2015, a 50%+ cumulative gap. This discount will persist until capex intensity drops or returns improve. Neither is likely in the near term.
Covenant risk. Telecom debt covenants typically include leverage ratios (net debt/EBITDA ≤ 3.0–3.5x) and interest coverage ratios (EBITDA/interest expense ≥ 3.5–4.0x). Current headroom is adequate (ratios at 2.8x and 4.2x, respectively, for Verizon). However, a 10–15% EBITDA decline (plausible under a recession scenario) would breach covenants. Breaches trigger renegotiation and higher cost of capital. Covenant violation risk is non-trivial for AT&T (2.9x leverage, 3.8x coverage).
Credit downgrade risk. All major telecom carriers are rated BBB or BBB+ (low investment grade). A single notch downgrade to BBB- moves them to sub-investment-grade status, triggering forced selling by index funds and institutional investors with credit-quality mandates. That selling would compress valuations a further 15–20%.
Exit Scenarios: Private Equity and Infrastructure Funds
Recognizing the capex trap, infrastructure funds and private equity have begun acquiring telecom assets and companies. The thesis: buy mature networks at a discount, cut capex (de-risk from 6G arms race), optimize operations, and harvest cash flow. Models assume 4–6% unlevered returns.
Tower and infra spin-offs. AT&T, Verizon, and Deutsche Telekom have spun or plan to spin tower companies (American Tower owns ~40K towers; Verizon Towers; DT Towers). This de-risks balance sheets and frees capex for cash distribution, but reduces leverage benefits and increases fragmentation (tower companies charge access fees, compressing margins).
Spectrum sales. Carriers facing liquidity stress may sell spectrum licenses or bandwidth to well-capitalized rivals or private networks. NTT Docomo considered this in 2024. Revenue from a spectrum sale (est. $5–15 billion, depending on spectrum and duration) would provide one-time relief but sacrifice long-term optionality.
Wholesale-only pivots. Some smaller carriers (e.g., Sprint → T-Mobile consolidation) have opted to become wholesale providers (selling network capacity to MVNOs) rather than competing in retail. This reduces capex but also commoditizes service and margins. Not viable at scale for major carriers due to ARPU destruction.
Private equity take-private. A PE consortium could acquire a major carrier (e.g., Vodafone, Telecom Italia, Telenor) at a 30–40% discount to historical valuation (capturing the capex-trap discount), cut capex aggressively, and harvest 6–8% cash returns over 5–10 years. This is plausible for regional carriers but unlikely for US “Big Three” due to size and regulatory complexity.
Conclusion: Structural Capex Trap
The telecom industry is caught in a structural trap: decades of capex deployed into an environment of zero incremental consumer demand and failed enterprise use cases. Operating costs rise faster than revenue. Refinancing risk looms as rates remain elevated. Equity returns lag; credit quality is brittle.
For investors: (1) avoid equity; (2) treat debt as carry with embedded option risk (downgrade or covenant breach); (3) consider infrastructure funds acquiring tower/fiber as alternative to carrier equity; (4) monitor dividend sustainability (AT&T > Verizon > German carriers > Japanese carriers in order of downside risk).
The next five years will reveal whether carriers can rationalize capex (6G deferral, Docomo dividend cut, Verizon debt reduction). Until then, stranded capex and operating cost drift will continue to compress returns.