Do Sensitivity Analysis - How Much Eps Is
EPS Sensitivity — How Much GAAP EPS Is At Risk Per Million Internet Subs Lost
Bottom Line
At the central calibration — $70.72/month residential broadband ARPU, 80% drop-through, FY2025 effective tax (22.7%) and A/N NCI (13.5%), 137.7M diluted shares — Charter gives up roughly $3.30 of GAAP EPS for every 1M Internet subscribers lost on a full-year run-rate basis. The defensible band is $2.50–$4.50/share per 1M subs, with drop-through assumption (70%–90%) and ARPU framing (broadband line vs. customer relationship) accounting for almost all of the spread. The same 1M-sub loss reprices to roughly $2.30/share post-Cox close (197M FDSO denominator) — so the EPS-loss arithmetic is materially less punitive once the deal closes, even before Cox's incremental EBITDA. Five million Internet sub losses (a ~17% impairment of the 29.7M Internet base, well outside historical experience) would compress FY2025-equivalent EPS from $36.21 to roughly $19.70 — a stress test, not a base case.
Central case: $/EPS per -1M Internet subs
Low end of defensible band
High end of defensible band
FY2025 actual diluted EPS
FY2026 consensus EPS (16-an. avg)
FY25 diluted shares (M)
The conversion ratio in one sentence. $3.30 of GAAP EPS per 1M Internet subs lost at the central case (broadband ARPU × 80% drop-through × 77.3% after-tax × 86.5% after-NCI ÷ 137.7M shares). Hold one assumption fixed, change another, and the answer moves $0.50–$1.20 — making the band as analytically important as the point estimate.
1. The Mechanical Bridge — EBITDA Down To EPS, One Million Subs At A Time
The drop-through is layered. Each line is a pre-set assumption the reader can interrogate.
The takeaway: of every $1 of revenue lost to a departing Internet sub, roughly 53 cents lands as net income to Charter shareholders. The other 47 cents leaks through (a) variable cost saves at the EBITDA line (20%), (b) federal/state tax (~12%), and (c) A/N's noncontrolling slice (~7%). D&A and interest are sunk and do not move with subs — the most common modeling mistake in this space is to flow them down with revenue, which materially overstates the EPS hit.
2. The Headline Grid — Sub Loss × Drop-Through
The spine of the tab. Rows are sub-loss scenarios (0M to -5M Internet subs); columns are drop-through assumption. Each cell is dollars of GAAP EPS lost, plus implied pro-forma EPS off the FY2025 actual base ($36.21) and the FY2026 consensus base ($41.91, avg of 16 analysts).
How to read this. At Charter's actual FY2025 print (-393K residential Internet subs, close to the -0.4M column), the EPS impact at 80% drop-through is roughly $1.30/share — well inside reported earnings volatility. At 1M, the hit is $3.30/share at the central case — ~9% of FY2025 EPS, ~8% of FY2026 consensus. At 3M (a ~10% Internet base impairment, comparable to the worst cable cumulative losses since 2022), the hit is $9.89/share — knocks pro-forma FY2026 EPS to $32.02, below Goldman's $185 PT-implied EPS but above Wells Fargo's $200 target math. At the 5M tail (a structural broadband break), pro-forma EPS lands below $20/share, multiple compression follows, and the value-trap thesis becomes the base case.
3. ARPU Framing — Is A "Sub Loss" The Internet Line Or The Whole Household?
The most important assumption after drop-through is what "subscriber" means. Charter discloses two anchors that bracket the right number; its own FY2025 revenue bridge sits between them.
The reconciliation. Industry convention treats the Internet line as the relationship anchor — losing Internet usually means losing the relationship — which pushes the right ARPU figure closer to the $119 relationship number than the $70 broadband-only number. But the offsetting move is in drop-through: a household loss drags video programming saves (dollar-for-dollar) and mobile MVNO wholesale costs down with it, so the drop-through compresses from ~80% (Internet-only) to ~55% (whole-household). The two effects roughly cancel — which is why all three framings cluster between $3.30 and $3.82 per share at central drop-throughs. The honest read is that the point estimate is robust to ARPU framing; the band width comes almost entirely from the drop-through choice.
4. Calibration Check — Does The Bridge Reproduce Charter's FY2025 Disclosure?
Calibration note. Charter's own MDA bridge attributes $380M of Internet revenue decline to 393K of residential Internet sub loss — an implicit $967/yr per departing sub, ~14% richer than the spot $848.6/yr broadband ARPU. The gap likely reflects bundle revenue allocation back to the Internet line (Internet revenue includes a portion of bundled discount that gets re-attributed when the customer cancels). Using the $967/yr anchor instead of the $848.6/yr broadband ARPU pushes the per-1M-sub EPS hit from $3.30 to $3.76 at 80% drop-through — still well inside the $2.50–$4.50 band.
5. Share Count Regime Overlay — The Second-Biggest Swing Factor
Per-1M-sub EPS sensitivity is mechanically inversely proportional to share count. Charter sits on the cusp of three different denominators, and which one applies depends entirely on whether the Cox transaction closes inside the Sept. 15, 2026 DOJ deadline.
The Cox close roughly halves per-share sensitivity — but it is not a free pass. Post-Cox FDSO of 197M is 43% larger than FY2025 diluted, so the same $454M NI hit dilutes to $2.30/share vs $3.30/share. Crucially, Cox also brings ~$5B+ of incremental EBITDA plus the $413M/year preferred dividend ($6B notional × 6.875%) paid out before common EPS — so the EPS baseline shifts too, not just the sensitivity coefficient. The clean way to think about this: this tab quantifies the Charter standalone broadband downside on a Charter standalone share count. Post-Cox math is a separate exercise that the catalysts and long-term thesis tabs handle.
6. The Two-Axis Heatmap — Sub Loss × Drop-Through, Broadband ARPU Framing
Visual restatement of the headline grid as a heatmap; same numbers, easier comparison.
7. The Honest Caveat — Don't Double-Count With ARPU Compression
The sensitivity above assumes subscriber loss in isolation, with ARPU held flat on the remaining base. That is an artificial condition: the same FWA / fiber pressure that drives sub losses is also what produced Charter's first negative residential ARPU print (-1.4% YoY) in 20+ years in Q1 2026. A coherent stress scenario should layer the two.
100 bps of sustained ARPU compression on the residual ~$23.8B residential broadband base is worth ~$0.92/share — roughly equivalent to a 280K-sub loss in EPS terms. The Q1 2026 -1.4% ARPU print alone, if it persists, is worth ~$1.29/share of annualized EPS drag, on top of any volume losses. A bear scenario that pairs -2M subs and -2% ARPU is roughly $8.4/share of EPS at risk — not the $6.59 the subscriber grid alone implies. The grid is therefore a conservative reference point, not a worst-case calibration.
8. Mobile Attach Drag — A Footnote Upward Adjustment
Spectrum Mobile is gated to having Spectrum Internet (per FY2025 10-K). Roughly 40% of broadband subscribers carry a Spectrum Mobile line; average ~1.4 lines per attached customer; mobile ARPU ~$32/month. Each Internet sub lost therefore takes its proportional share of attached mobile revenue with it — but mobile contribution margin is lower (~25–35%) because of the Verizon MVNO wholesale fee.
Net effect: roughly $0.30/share additional EPS drag per 1M Internet subs lost from second-round mobile attrition — about a 9% upward adjustment to the headline $3.30 figure. Not large enough to change the band, but a real number to keep in mind when modeling later years (FY27+) when mobile penetration is higher.
9. Reference Frame — Where Each Scenario Sits Versus Historicals and Estimates
10. What This Tab Does Not Try To Answer
This is a partial-equilibrium, static, full-year run-rate bridge. It deliberately holds five things fixed that, in the real world, would all move together:
This tab quantifies how much EPS is at risk; it does not assert which scenario is the base case. The forward-estimates-claude.md and compare-how-big-of-a-threat-is-fixed-claude.md tabs own the question of whether Charter will lose 1M / 3M / 5M subs. This tab tells the PM the conversion ratio: if the bear thesis is right by 1M subs, $3.30/share of EPS is the marker; if by 3M, $9.89. The single highest-leverage watch item is the Q2 2026 earnings print on July 24, 2026 — both Internet net adds and residential ARPU are reported on the same day.
11. Final Decision Frame
The bridge is intentionally simple: drop-through, tax, NCI, share count — in that order of importance. A PM who walks away with the central conversion ratio ($3.30/share per 1M subs at 80% drop-through, FY25 shares) and a sense of the band ($2.50–$4.50) can convert any subscriber-loss scenario into an EPS underwrite in real time. The honest read is that the equity at $140 is pricing roughly 2.5M–4M of Internet sub losses in EPS-only terms (pro-forma FY26 EPS of $26–$32 against a ~5× P/E gives $130–$160 fair value); the sensitivity grid above lets the PM judge whether that's an over- or under-reaction once Q2 prints on July 24.