MAJORHOLDERS.COM RESEARCH
Industry Research · Broadband Infrastructure

Starlink's Capacity Curve: From Rural Niche to Broadband Contender

SpaceX's constellation ended 2025 near 450 Tbps of total capacity. With Starship-launched V3 satellites adding ~60 Tbps per flight, we model the network's trajectory through 2035 — and what it means for Comcast, Charter, AT&T, Verizon and T-Mobile.

~450 Tbps
Constellation capacity, end-2025
60 Tbps
Added per Starship V3 launch
12M+
Global subscribers, mid-2026
~13 Pbps
Base-case capacity by 2034

1. Where capacity stands today

Starlink's network capacity reached roughly 445–450 Tbps by late 2025, up from about 40 Tbps at the end of 2022 — a 10x expansion in three years. The company currently deploys about 5 Tbps of new capacity per week using Falcon 9 launches of second-generation (V2 Mini) satellites; in 2025 alone, more than 3,000 V2 Minis added roughly 270 Tbps.

The step change arrives with the third generation. Each V3 satellite is designed for over 1 Tbps of downlink (10x a V2 Mini) and 200+ Gbps of uplink (24x), with a single Starship launch projected to add ~60 Tbps — more than 20 times what a Falcon 9 mission adds today. SpaceX targeted the first half of 2026 for initial V3 launches. Separately, new FCC power-flux rules adopted in spring 2026 allow up to eight satellites to serve the same area simultaneously (versus effectively one before), a potential ~7x boost in usable capacity over the U.S. without launching anything.

The key constraint: Starlink's capacity is a shared, planet-wide pool. At ~450 Tbps across 12M+ subscribers, average provisioned capacity is on the order of ~35–50 Mbps per user — the reason congested U.S. cells still carry demand surcharges. Analysts estimate the current network supports ~20 million broadband subscribers globally.

2. Capacity projection, 2026–2035

We model three scenarios driven almost entirely by one variable: Starship launch cadence dedicated to Starlink V3. Assumptions: V2 Mini launches (~5 Tbps/week) taper as Starship ramps; each Starship flight carries ~60 V3 satellites (~60 Tbps); satellites deorbit after ~5 years, so retirements increasingly offset gross additions late in the window. The base case is calibrated to New Street Research's independent assessment of ~20,000 V3-class satellites and ~29x capacity growth from 2025 to 2034.

Total Starlink network capacity, 2022–2035 (log scale)
100 Tbps 1 Pbps 10 Pbps '22'24 '26'28 '30'32 '34 PROJECTED → 450 ~950 ~7.3k ~13.2k
Actual (Starlink disclosures) Base case Conservative Aggressive
Values in Tbps, end of year, net of satellite retirements. Historical figures from Starlink network updates; projections are majorholders.com illustrative estimates.

Conservative

~8 Pbps by 2035

Starship suffers continued delays; Falcon 9 carries the load through 2028 with V3 flights ramping to only ~15–20/yr by 2030. Supports ~90–120M subscribers.

Base case

~13 Pbps by 2034

Starship reaches ~15 Starlink flights in 2027, ~40–45/yr by 2030. Cumulative ~18–20k V3 satellites by 2034; ~29x capacity vs. 2025. Supports ~200M subscribers.

Aggressive

~28 Pbps by 2035

Consistent with SpaceX's stated ambition of ~400 Starship flights in four years, a majority devoted to Starlink, plus a higher-capacity V4 late in the window.

Base-case launch and capacity model, year by year

Annual launch cadence → capacity build (base case)
YearFalcon 9 flightsStarship flightsSatellites added Gross Tbps addedTbps retiredNet capacity (Tbps)Subs supported*
2023 (actual)~630~1,900~55~5~90~4M
2024 (actual)~890~2,300~100~10~180~8M
2025 (actual)~100+0~3,000~270~15~450~20M
2026E~90~5~2,450+560−60~950~42M
2027E~60~15~2,350+1,070−70~1,950~85M
2028E~35~25~2,350+1,590−140~3,400~150M
2029E~15~35~2,450+2,140−240~5,300~200M+
2030E~5~40~2,500+2,420−420~7,300~200M+
2031E0~45~2,700+2,700−1,000~9,000>200M
2032E0~45~2,700+2,700−1,000~10,700>200M
2033E0~45~2,700+2,700−1,200~12,200>200M
2034E0~45~2,700+2,700−1,700~13,200>200M
2035E0~45–50 (V3/V4)~2,800+3,000−1,800~14,500–16,000>200M

*Subscribers supported at current provisioning ratios (~44,000 subs per Tbps, derived from ~20M subs on ~450 Tbps). Actual serviceable subscribers depend on geographic distribution of demand vs. beam capacity — the constellation cannot concentrate all capacity over one country. Falcon 9 carries ~24–28 V2 Minis per flight (~2.5–3 Tbps); Starship carries ~60 V3s (~60 Tbps). Satellites retire after ~5 years, so V2 Minis launched 2023–2026 roll off 2028–2031, and early V3s begin retiring in 2031 — the drag visible in later years.

3. The terrestrial comparison: why "total Tbps" flatters satellites

Comcast, Charter, AT&T, Verizon and T-Mobile do not publish a single network-capacity figure, because wired networks scale locally. A single modern fiber pair carries multiple terabits; a fiber ISP provisions ~10 Gbps to a group of ~32 homes and simply adds equipment where demand grows. Comcast has historically doubled network capacity roughly every 18 months, engineering for peak-hour traffic and planning 12–24 months ahead of ~40–45% annual usage growth. In effect, terrestrial capacity is unbounded per neighborhood; Starlink's is a fixed budget per square mile of sky.

The fair comparison is therefore subscribers and share, where Starlink remains small in the U.S. but is the fastest-growing platform globally (2.3M subs in 2023 → 4.4M in 2024 → 8.9M in 2025 → 12M+ in mid-2026).

U.S. broadband subscribers by provider, Q1–Q2 2026 (millions)
Charter (Spectrum)
~30.0
Comcast (Xfinity)
~29.5
Verizon (Fios+Frontier+FWA)
~16.8
AT&T (Fiber+Lumen+AIA)
~15.0
T-Mobile (FWA+fiber)
~8.0E
Starlink (U.S.)
~2.5E
Approximate figures compiled from Q1 2026 company disclosures and analyst reports; T-Mobile did not disclose an FWA count in Q1 2026 and Starlink does not report U.S. subscribers — both are estimates (E). Charter figure excludes the pending Cox acquisition (~6M subs, expected to close Aug–Sep 2026).

Positioning snapshot

Cable is the share donor. Both Charter and Comcast peaked in 2023 (32.3M and 30.6M subscribers) and have each lost over a million since. Q1 2026: Comcast −65k (its best quarter in two years after simplified five-year pricing), Charter −120k. Switching panels show cable brands with deeply negative net switching rates, while fixed wireless was the disruptor: the Big 3 carriers now serve 20M+ FWA subscribers with spectrum capacity for up to 32M. Fiber wins where it exists — AT&T reached 37M fiber passings (targeting 60M by 2030) after acquiring Lumen's mass-market assets, and Verizon closed its $20B Frontier deal in January 2026, pushing fiber connections to ~10.8M. Consolidation is accelerating on both sides: Cox–Charter approved, a Comcast–Charter combination openly debated by analysts.

4. How Starlink's capacity curve changes the dynamic

2026–2028: from waitlists to open signups

Roughly doubling capacity in 2026 (to ~950 Tbps base case) plus the FCC's 8-satellite EPFD reform lets Starlink retire congestion surcharges, open capped suburban/exurban markets, and market gigabit-tier service. Promotional pricing ($35/mo introductory residential offers) makes it a direct FWA competitor for price-sensitive households — pressuring T-Mobile and Verizon's FWA growth at the margin and giving cable yet another low-price alternative to defend against.

2029–2031: a national value tier

At 5–9 Pbps, Starlink's U.S.-allocable capacity could plausibly support 8–15M domestic subscribers at cable-like typical speeds. That's enough to make it a top-5 U.S. ISP by subscriber count and a permanent pricing check on every terrestrial player — while still structurally unable to match fiber's per-home capacity in dense metros. Expect its share gains to come disproportionately from cable's rural/exurban footprint and from FWA churn, not from fiber.

2032–2035: ceiling defined by physics and demand, not supply

At 13+ Pbps and 200M+ global subscriber capability, growth constraints shift from the network to marketing, terminal economics, and terrestrial competition in developed markets — the same limits New Street Research flags in its 100M-subscriber-by-2034 forecast. The most durable outcome: fiber holds the premium tier, a merged cable sector defends the middle with converged mobile bundles, and Starlink + FWA commoditize the value tier while Starlink takes near-total share of the truly rural and mobile (RV, maritime, aviation, direct-to-cell) segments globally.

Key risk to every scenario: Starship. The vehicle has a bumpy flight history, and each year of delay defers ~1–2 Pbps of capacity. A secondary risk is Amazon's LEO constellation — well-funded but, at ~700 initial satellites supporting ~1.5M subscribers, several years behind on capacity.

5. Sources

  1. Starlink, "Network Update" — capacity, V3 specifications, launch cadence. starlink.com/updates/network-update
  2. Data Center Dynamics, "Starlink targets 2026 for terabit satellites for launch with Starship" (May 2026)
  3. CircleID, "Starlink Update: Expansion, Performance Gains and Network Developments" (2026)
  4. DISHYtech, "Starlink Just Had a Massive 2025 — and 2026 Could Be Even Bigger" (Jan 2026)
  5. Advanced Television / New Street Research, "Starlink could reach 100m subs by 2034" (May 2026)
  6. TMF Associates (Tim Farrar) consumer analysis via SmallSat News (June 2026) — subscriber history, ~20M-sub network capacity
  7. 5Gstore, "Starlink Capacity Gets 700% Boost From New FCC Rules" (May 2026)
  8. FactSet Insight, "U.S. Wireless Operators Increase Focus on Fiber" (July 2026) — cable/fiber/FWA subscriber trends
  9. Fierce Network — Q1 2026 earnings coverage (Verizon, AT&T, T-Mobile); New Street Research FWA capacity estimate
  10. The Motley Fool, "Where Are Comcast and Charter's Internet Customers Going?" (May 2026)
  11. Mobilewalla, "What Broadband Switching Data Reveals About Market Competition" (June 2026)
  12. Company filings and earnings releases: Comcast, Charter, AT&T (10-Q Q1 2026), Verizon, T-Mobile (8-K 2026)
Disclosures. Projections in Sections 2 and 4 are illustrative estimates constructed by majorholders.com from public disclosures and third-party analyst assessments; SpaceX does not publish multi-year capacity guidance. "Subscribers supported" reflects network capacity at current provisioning ratios, not a demand forecast. This report is for informational purposes only and is not investment advice, an offer, or a solicitation. Figures marked E are estimates. Data as of July 2026.