Reasons

Reasons Capex Is So High Every Year

Charter is in a once-in-a-decade overlap of three large physical projects — a federally subsidized rural network build, a 1.2/1.8 GHz "network evolution" DOCSIS 4.0 rebuild, and continued line-extension expansion — layered on top of an already capex-heavy cable operating model that has to renew $2B+ of customer equipment and $2B of support assets every single year. FY2025 capex was $11.7B (21.3% of revenue), versus $7.6B in FY2021 (14.8% of revenue). The spike is driven almost entirely by line extensions (rural + non-subsidized) and the upgrade/rebuild line; the recurring buckets are roughly flat. Management has guided FY2026 capex to ~$11.4B and framed the post-2027 run-rate as materially lower, with the rural program targeting completion at end-2027 and network evolution scheduled to finish in 2027.

Source for category dollars and FY2026 guide: FY2025 10-K MD&A "Capital Expenditures" section. Capex/D&A computed from cash_flow.json and income.json.

1. The Long-Run Picture: Two Different Capex Regimes

For roughly five years (FY2019–FY2021) Charter ran at $7.2–$7.6B of capex, or 14.8–15.7% of revenue. Beginning in FY2022 the company stepped that base up by ~$4B per year — a step that is still in place today and is the entire reason free cash flow has compressed from a $8.6B peak in FY2021 to $4.4B in FY2025.

The earlier (2017–2018) bulge is the post–Time Warner Cable / Bright House integration capex on the merged footprint — a different, one-time event. The current bulge has a different cause: three concurrent physical projects (rural build, network evolution, expansion line extensions) that did not exist together in any prior year. Revenue stalled around $54–55B across FY2022–FY2025, so the capex step-up shows up undiluted in the ratio: 14.8% → 21.3% in four years.

2. The Composition — Why the Spike Is Real, Not Inflation

The single most important reason capex is high every year is that two of the five NCTA categories — Line Extensions (where the subsidized rural buildout lives) and Upgrade/Rebuild (where the DOCSIS 4.0 "network evolution" lives) — have roughly tripled their FY2021 size, while the recurring categories (CPE, scalable infrastructure, support capital) are roughly flat.

FY2021 reflects the original FY2021 10-K NCTA table, which presented Line Extensions as a single line. The subsidized rural construction sub-line was first separately disclosed in the FY2022 10-K MD&A (showing FY2022 and the restated FY2022 mix used here). The FY2022 column above uses the restated FY2023 10-K prior-year column for comparability.

Four readings from the table. (1) Of the $4.0B step-up from FY2021 to FY2025, roughly $2.2B is the subsidized rural construction program — a bucket that effectively did not exist at scale four years ago. (2) Another $1.2B of the step-up is in Upgrade/Rebuild, which the 10-K explicitly says "includes our network evolution initiative" (the DOCSIS 4.0 / 1.2–1.8 GHz rebuild). (3) CPE has been roughly flat at $2.2B even as customers declined — confirming this is a hardware-replacement bucket that does not scale down with subs. (4) Support capital grew $343M from FY2021 to FY2025 and was the largest single driver of the YoY increase from FY2024 to FY2025, alongside higher network-evolution spend (per the FY2025 10-K MD&A narrative).

3. Reason-by-Reason Walkthrough

Each NCTA category is large for a different physical or operational reason. Below is each bucket's FY2025 dollar amount, what is actually being built, why it costs what it costs, and what would change it.

Bucket dollars: FY2025 10-K MD&A "Capital Expenditures" section. Network-evolution timing and rural completion targets: Q4 FY2025 earnings release and Q1 FY2026 earnings release; rural passings count of 483K in FY2025: Q4 FY2025 earnings release. Competitive context (~43% combined FTTH overbuild reach): review/final/competition-claude.md.

4. Structural vs. Cycle-Driven: How Much of the $11.7B Will Stay?

This is the key question for the equity. The FY2025 NCTA buckets split cleanly into a structural recurring floor that will roughly persist after rural and network evolution finish, and a cycle-driven peak that is, by definition, finite.

The ~$5.4B structural floor is broadly consistent with the FY2019-FY2021 pre-spike base: in FY2021, CPE + Support + Scalable totaled $5.3B and the LE/Upgrade buckets together were $2.3B. Adjusting for inflation and for the WiFi 7 / mobile-device cycle the structural floor today is likely closer to $5.5-6.5B, with the rest fading as programs close. That matches the company's own framing that post-2027 capex should run materially below FY2025 levels (the long-term thesis tab cites a sub-$8B target post-2027).

5. Capex Is Now Running Faster Than the Asset Base Depreciates

A second reason capex is "high" is mechanical: at $11.7B against $8.7B of depreciation, Charter is adding to its plant faster than the existing plant is being expensed. The capex-to-D&A ratio inverted from 0.82x in FY2021 to 1.34x in FY2025 — meaning the gap between cash spent and book depreciation is what compresses FCF below GAAP net income, even though the income statement looks healthy.

There is also an accounting amplifier on top of the physical-build story. The forensics review notes Charter capitalizes $2.6B of direct labor and overhead in FY2025 — roughly $0.2B higher than FY2024 — based on internal time-and-motion standards. That practice is industry-standard for cable operators and has been stable, but it does mean that some of the "capex" line is recurring labor cost that lives on the balance sheet rather than the income statement. This is a real reason capex looks high relative to D&A, but it is not the cause of the FY2022–FY2025 spike, which is overwhelmingly physical (rural + network evolution). The capitalization-policy effect is a level effect; the cycle effect is the slope.

Source: review/final/forensics-claude.md, section 3 "Earnings Quality."

6. What Management Is Saying

Management has been unusually explicit that the current capex level is cycle-driven and that the path back to a lower run-rate is tied to defined program end-dates. Three statements from FY2024–FY2026 earnings releases capture the position:

"Our multi-year investments in network evolution, expansion and execution are delivering tangible results. By having the best network, the best products and delivering customers the most value with unmatched service, we are well-positioned for customer and profitability growth and have clear visibility to free cash flow growth following this unique one-time investment cycle."

— Chris Winfrey, President & CEO, Q4 FY2024 earnings release (Jan 31 2025)

"Spectrum is evolving its connectivity network to offer symmetrical and multi-gigabit Internet speeds across its entire footprint and has launched symmetrical Internet service in several markets. Unlike competitors, Spectrum upgrades its network to serve all of its passings and can do so at a much lower cost. Spectrum expects to complete its network evolution initiative in 2027."

— Charter Q4 FY2025 earnings release (issued Jan 30 2026)

"Charter currently expects full year 2026 capital expenditures to total approximately $11.4 billion. The actual amount of capital expenditures in 2026 will depend on a number of factors including … the pace of Charter's network evolution and expansion initiatives, supply chain timing and growth rates in Charter's residential and commercial businesses."

— Charter Q4 FY2025 earnings release (FY2026 capex guide)

The FY2026 guide of $11.4B — only $260M below FY2025 — is the proof point that capex stays high for one more year before the stepdown is visible. Management has not given an FY2027 capex number; the public framing is that program completion in 2027 enables a materially lower run-rate from FY2028 onward.

7. What Changes the Answer

The "why is capex high?" answer is true today and true through FY2026. The first proof point of a different answer is the FY2027 print.

Material limitations noted: (i) FY2021 NCTA bucket detail does not separately disclose subsidized-rural capex (the program was first separately disclosed in the FY2022 10-K), so the FY2021 column in section 2 attributes the full FY2021 line-extensions amount to "Other LE." (ii) The structural-vs-cycle split in section 4 includes a judgmental allocation of roughly $1.2B of FY2025 scalable infrastructure to the structural floor (vs. ~$0.3B to the DAA / network-evolution cycle) — the 10-K does not provide this sub-line split. (iii) No FY2027 capex guide has been disclosed; the post-2027 "sub-$8B" framing is taken from the long-term thesis review file, which sources it from public CEO commentary.