Uniti Group

Refresh delta (since 2026-06-20)

What changed

1. The growth crossover reversed. Q2 2026 revenue $909.7M, −5% pro forma YoY; Adjusted EBITDA $357.1M, −10% pro forma YoY and −19% sequentially from Q1's $441.6M. The prior dossier's central evidence — "the first time since the combination that the company posted year-over-year growth in both consolidated revenue and Adjusted EBITDA" — lasted exactly one quarter.

2. That Q1 inflection was substantially one lumpy deal. Pro-forma sales-type-lease revenue was $88.2M in 1H but only $16.8M in Q2 — i.e. ~$71.4M landed in Q1, from "the IRU dark fiber agreement with a large hyperscaler entered into in the first quarter of 2026". Underneath it, recurring service revenue fell 9% YoY in Q2 and 8% in 1H ($852.9M vs PF $936.5M; $1,741.9M vs PF $1,900.9M). Management now warns the Q4 repeat may not land: "there is a possibility that some of those deals could slip into early 2027" — Gunderman.

3. Double miss, and the tape finally punished it. EPS −$0.68 vs −$0.43 consensus (a second source says −$0.47); revenue $909.7M vs ~$916.2M consensus. The stock gapped +5.11% premarket to $10.90 then fully reversed to close −9.6% at ~$9.38. At $9.39 on 2026-08-10 it is −19% from the $11.64 in the prior dossier.

4. Guidance split down the middle: EBITDA up, bottom line materially worse. FY2026 revenue raised to $3.63–3.68B (from $3.61–3.66B) and Adjusted EBITDA to $1.45–1.50B (from $1.43–1.48B) — but net loss WIDENED to $(530)–$(480)M from $(450)–$(400)M, net interest expense raised to $805M (from $785M), and Kinetic capex raised $100M to ~$1.27B to pull 2027 construction forward.

5. A second Kinetic securitisation — and the debt stack still grew. Kinetic ABS 2026-2: $1,140.7M priced 2026-06-05 at ~6.180% WAC, closed 2026-07-15, ARD June 2033. $500M of proceeds went to senior secured paydown — $19.8M of term loan (lenders declined most of the $167.8M offered) plus asset-sale offers increased from $332.2M to $480.2M, expiring 2026-08-20. Net of it all, roughly $640M of new gross debt funds capex. Offsetting good news: blended debt yield now ~6.5%, down from ~12.5% in February 2023.

6. Leverage got worse and book equity is nearly gone. Net debt $10,072M at 6/30 vs $9,701M at 3/31 — debt principal was flat ($10,683.1M → $10,680.6M) but cash fell $373.7M ($982.6M → $608.9M), so the build burned the Q1 refinancing cash. Total shareholders' equity $161.5M, down from $319.7M at 3/31 and $380.4M at YE2025.

7. Structural subordination is being manufactured, quarter by quarter. Unrestricted (ABS) subsidiaries now hold $2,007.7M of assets (15.5% of total, up from $736.4M / 6.1% at YE2025) and $1,764.5M of debt (from $822.1M), and produced 45% of consolidated operating income on 11% of revenue in 1H. Post-July the Kinetic ABS box holds fibre in ten states. These entities are expressly outside the restricted group: "Neither the Company nor any subsidiary… will guarantee or in any way be liable for the obligations of the Kinetic ABS 2026-2 Obligors".

8. The prior dossier's maturity wall was wrong, in the company's favour. It placed "$2.2B 8.25%" notes in 2028. The 10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited. is explicit: the 8.25% secured notes mature October 1, 2031. The real ladder at 6/30/26: 2026 $10.0M · 2027 $886.5M · 2028 $1,120.0M · 2029 $1,299.0M · 2030 $1,220.1M · thereafter $6,145.0M. There is no 2028 cliff. Correcting this materially softens the single loudest bear point in the prior file.

9. Genuinely record operations underneath the bad P&L. Q2: 141,000 fibre passings, 38,000 net fibre adds (record), 603,000 fibre subs +25% YoY, ~2.1M homes passed at 29% penetration; Fiber Infrastructure record bookings of $2.2M MRR, +30% over the prior record, with >50% Waves/lit rather than dark fibre and 37% blended anchor lease-up cash yields, "the highest we've ever seen". BEAD is now quantified: provisional awards in seven states totalling $184.3M for ~58,000 locations, all seven NTIA-approved, two of seven contracts executed (AR + TX, $9.8M / 3,800 locations).

What held

The structural thesis stands, but one of its two legs has visibly buckled.

• Holds: the asset is real and the fibre engine is accelerating, not stalling — record passings, record net adds, record wholesale bookings, 29% penetration. The capital-structure repair is real and better than the prior dossier credited: blended yield 6.5% vs 12.5% three years ago, no 2028 cliff, ~$3B of securitisation raised, covenant-compliant, and liquidity of $608.9M cash plus $838.1M of revolver capacity. Customer concentration remains solved (no >10% customer). Gunderman's strategy is unchanged and he is executing it.
• Buckled: the prior dossier's operative claim was that "the central operational race is fiber growth out-running legacy erosion, and as of Q1 2026 it crossed over." It did not cross over. Kinetic total revenue is −7% PF, Uniti Solutions −16% PF, and only Fiber Infrastructure (+10% PF) grows — and even that is flattered by $6.1M of one-time early-termination penalties and a favourable billing-dispute reserve release. Management's own framing is now "fiber-based revenue is on track to represent the majority of total revenue by the end of 2026" — a mix milestone, not a growth milestone.
• The arithmetic that matters most: the prior dossier annualised Q1's $441.6M to $1.77B and computed ~5.7x net leverage. Q1 contained ~$71.4M of non-repeating IRU revenue at near-full incremental margin. Strip it from 1H and the clean run-rate is ~$720M × 2 ≈ $1.44B. On that denominator net leverage is ~7.0x, not 5.7x. Management's own raised guidance corroborates it: $1.475B midpoint less $798.7M of 1H actual implies a 2H of $676.3M, or ~$338M/quarter — below the $357.1M Q2 just printed. The company raised EBITDA guidance to a number that still implies the second half runs beneath the second quarter.

Competitive Advantages — refreshed (partial; new competitive data)

The moat thesis is unchanged in shape but now has numbers on both sides, and they cut both ways.

Better than the prior dossier assumed: Kinetic's overbuild exposure is genuinely low. John Harrobin (President, Kinetic): "we're fortunate in a sense that we only compete with cable in less than 60% of our Fiber territory" versus "mid-80s to low 90s" for peers. On the wholesale side the moat is monetising unusually well — 80% of hyperscaler business "includes selling all or at least partial existing infrastructure," producing 37% blended anchor lease-up cash yields. Selling capacity on already-buried fibre at a 37% cash yield is the strongest single economic fact in this refresh.

Worse than the prior dossier assumed: the prior file hypothesised that "where AT&T/Comcast or FWA arrive, pricing power compresses." That is no longer a hypothesis. Consumer ARPU is falling, and management attributes it to "the competitive promotions that cable started with and the telcos matched on" — expecting ARPU to return to only "low single digits" growth from Q4 2026. Uniti is a promotion-taker, not a price-setter, even inside its own incumbent footprint. A 29% penetration rate on 2.1M homes passed is respectable but not commanding.

Net: the moat is real where the fibre is buried and there is no cable competitor, and it is thin everywhere else. Unchanged conclusion, sharper evidence.

Segments — refreshed (mix moved materially)

Q2 2026 GAAP vs pro-forma Q2 2025:

The mix story has inverted. The prior dossier framed Kinetic as "the growth engine" and Fiber Infrastructure as the declining legacy landlord book. As of Q2 2026 the reverse is true on a pro-forma basis: Fiber Infrastructure is the only segment growing, and Kinetic is shrinking in aggregate. Consumer fibre inside Kinetic is growing fine; the ILEC wrapper around it is not.

Quality-of-earnings note that belongs here as much as in Lens 10: Kinetic's compensation expense fell 20% "primarily reflect[ing] the beneficial effects of an increase in capitalized internal labor costs". Labour moved from opex to capex flatters contribution margin and Adjusted EBITDA while inflating capex. The 42% Kinetic margin is not comparable to a prior-year margin computed before that shift.

Earnings Result — Q2 2026 (reported 2026-07-30)

Versus consensus: a double miss. EPS −$0.68 vs −$0.43 (a second source reports −$0.47 — the miss is $0.21–0.25 either way); revenue $909.7M vs $916.2M.

Guidance (raised on top, widened underneath):

The company describes this as updating "primarily for business unit level revisions, the recently completed Kinetic asset securitization, and transaction related and other costs".

Balance-sheet and cash-flow flags — this is where the quarter is worst:

• Q2 operating cash flow was $78.5M against $195.6M of interest expense. The business did not generate enough cash from operations in Q2 to pay its own interest. The prior dossier's finding that the business doesn't earn its interest at the operating-income line has now propagated down to the cash line.
• Q2 capex $450.0M exceeded Q2 Adjusted EBITDA of $357.1M — 126% of EBITDA reinvested, up from 79% in Q1.
• 1H levered FCF −$459.8M (OCF $339.4M − capex $799.2M), before another $34.8M of grant-funded capex.
• Total shareholders' equity $161.5M, from $380.4M at YE2025. Accumulated deficit widening; a further two quarters at the Q2 loss rate takes reported equity through zero.
• Net debt $10,072M. Liquidity $608.9M cash + $838.1M revolver capacity = ~$1.45B, down from ~$1.84B at Q1.
• Purchase price allocation finalised in Q2 with only a $0.2M measurement-period adjustment; goodwill $1,158.5M unchanged.

Market reaction. Gapped +5.11% premarket to $10.90 on the record fibre metrics and raised guidance, then reversed all day to close −9.6% at ~$9.376, having traded as low as $9.40 intraday. Research-layer conflict surfaced, not resolved: these two sources disagree in sign because they measure different moments; the close is the fact that matters, and the close was a rout. This is the first print since the merger where the market refused to look past the GAAP loss — a direct contradiction of the prior dossier's Lens 8 conclusion that the tape "correctly learned to ignore" GAAP EPS for this name.

Earnings Calls — refreshed (sentiment shift)

The transcript is now on disk, so this lens upgrades from `` summary to primary.

The tone shifted from "the thesis is proven" (Feb 2026) to "2026 is an inflection year" (July 2026) — and that is a demotion, not a promotion. Gunderman in February: "the original thesis has proven to be true." Gunderman in July: "2026 is an important inflection year for Uniti and the terrific industry tailwinds we just discussed are fueling that pivot." A company that has proven its thesis does not need to relabel the year an inflection year. The claim moved from past tense to future tense while the numbers moved backwards.

What management now leads with: volume metrics, not financial ones — passings, net adds, bookings MRR, penetration, cash yields. All genuinely records. What it no longer leads with: the revenue/EBITDA growth crossover it led with in Q1.

New vocabulary, and it is the datacenters beat arriving in the story: "The use of Agentic AI is now upon us and growing… the time line and TAM appear ahead of our prior expectations"; "Over 50% of the new bookings this quarter were Waves or lit capacity as opposed to dark fiber, reinforcing the pivot from the build cycle to more lease-up and inference"; Waves funnels representing "approximately 1.3 petabytes of traffic from mostly new customers". This is the first call where inference demand is a named revenue driver rather than a background tailwind.

Credibility-positive admissions (consistent with the prior dossier's read that this team concedes rather than spins): Harrobin owning the ARPU decline as a response to "competitive promotions that cable started with and the telcos matched on"; Gunderman flagging before it happens that Q4 dark-fibre deals "could slip into early 2027"; Harrobin conceding "there's going to be a slight increase in our cost of Fiber in the outer years, '27 and beyond."

Credibility-negative: framing a $100M capex increase as a "pull-forward of 2027 spending to accelerate growth" is the single most convenient available explanation for spending more, and it is unfalsifiable until 2027 capex actually comes in lower. Watch it.

New and important: Bullington identified $500M–$1B of non-core assets for monetisation over 12–36 months, adding that "the monetization of these assets would have a negligible effect on our adjusted EBITDA as many of them are underutilized today and currently produce minimal to no cash flow". The 10-Q echoes it: "the Company may also seek opportunities to monetize certain non-strategic assets". If the assets truly produce no cash flow, this is free deleveraging — the best available lever and the one to track hardest.

Comps — partial refresh (UNIT's own multiple only)

Peer multiples for LUMN / FYBR / CCOI were not sourced in this run either and remain n/a. What changed is UNIT's own mark, and it changed in a direction the share price alone would not suggest.

The stock fell 19% and got more expensive. The prior multiple was computed by annualising a quarter that contained ~$71.4M of non-repeating hyperscaler IRU revenue; on the company's own raised full-year guidance the multiple is ~8.8x, and on the clean ~$1.44B run-rate it is ~9.0x. That is not a distressed multiple for a business whose recurring service revenue is declining 9% and whose levered free cash flowFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. is deeply negative. The prior dossier's conclusion that "the valuation is not the edge" survives; its implication that the name was "roughly fair-to-cheap" does not.

Sell-side: consensus Hold; median PT $11.875 across six analysts; average PT $11.14 — sources conflict, both cited. JPMorgan cut $12 → $11, Neutral, 2026-07-31.

Stock-Price Catalysts — refreshed

New >5% moves and catalysts since the boundary:

• 2026-07-30: −9.6% on the Q2 print (from $10.37 to ~$9.376), after a +5.11% premarket gap that fully reversed. Driver: the double miss and the 10% pro-forma EBITDA decline, which the market chose over the record operational metrics and the raised EBITDA guide.
• Boundary → print: roughly −11% ($11.64 on 2026-06-18 → $10.37 on 2026-07-29) with no single identified catalyst; drift.
• 2026-06-05: Kinetic ABS 2026-2 priced — $1,140.7M at 6.18% WAC. A financing-cost datapoint, which for this name is the equity.
• 2026-07-15/23/31: the asset-sale-offer sequence — $500M of ABS proceeds routed to senior secured paydown; term-loan lenders declined $148M of the $167.8M offered, pushing the note offer up to $480.2M, expiring 2026-08-20. Lenders preferring to hold rather than be prepaid is a mild credit-confidence signal.

What the market now reacts to — revised. The prior dossier concluded the market rewards Adjusted EBITDA and ignores GAAP EPS. Q2 falsified that. Record passings, record net adds, record bookings, a raised EBITDA guide and a 600bp improvement in blended debt cost were all in the release, and the stock still fell 9.6%. What moved it was the direction of EBITDA (−10% PF) and the widened net-loss guide, not its level. The operative rule is now: the tape will forgive a large loss, but not a shrinking EBITDA.

Forward catalysts: Q3 print (~late Oct 2026) — the clean test of whether service revenue stabilises without an IRU; the 2026-08-20 asset-sale-offer expiry; execution (or not) of the $500M–$1B non-core monetisation; the five unexecuted BEAD contracts; the 2027 maturity ($886.5M, including the $306.5M convertible).

Management — refreshed (capital allocation only)

No management changes found; Gunderman (CEO), Bullington (CFO) and Harrobin (President, Kinetic) all presented Q2. Bios, Elliott's ~25.2% stake, board-designation rights, the corporate-opportunity waiver and the 11%→16% escalating preferred are unchanged — see the previous dossier, Lens 9.

What changed is the capital-allocation playbook, and it is now a third thing. The prior dossier described it as: suspend the dividend, plow everything into FTTH, refinance lower and longer, delever via EBITDA growth. Q2 added a fourth leg that was not in the prior file: carve the best assets into bankruptcy-remote unrestricted subsidiaries, securitise them, and use the proceeds to fund the build — with non-core asset sales ($500M–$1B) as the deleveraging mechanism instead of EBITDA growth. That is a meaningful admission. Deleveraging via EBITDA growth requires EBITDA to grow; it fell 10%. Deleveraging via asset sales does not.

Judgement: this is competent, creative liability management by a team that has done it before — the 600bp yield improvement is a genuine achievement and the ABS market is clearly open to them. It is also, unavoidably, the equity financing its own growth by pledging the collateral that backs it. Both readings are true and the tension is the investment case.

Forensic Red Flags — refreshed

Carried from the prior dossier and still live: the FY2025 $1,683.9M non-cash lease-settlement gain contaminating TTM screens; reverse-merger comparability; goodwill $1,158.5M (with $515.5M of accumulated impairment already sitting against Fiber Infrastructure — a precedent); ICFR first-year exclusion of Windstream.

New in Q2 2026:

1. Revenue quality inside the one growing segment. Fiber Infrastructure's +10% contains $6.1M of one-time early contract termination penalties and a Uniti Wholesale increase attributed partly to "favourable adjustments to our reserves maintained for billing disputes with other carriers". Penalty income and reserve releases are not run-rate. Strip the penalties and Uniti Fiber is roughly flat.
2. Capitalised internal labour is flattering EBITDA. Kinetic compensation fell 20% "primarily reflect[ing] the beneficial effects of an increase in capitalized internal labor costs". Cost moves from the income statement to the balance sheet; margin and Adjusted EBITDA improve; capex rises. Disclosed, legitimate, and quantitatively unquantified — the filing gives no dollar figure for the shift, so the true like-for-like margin is not computable from the disclosure.
3. Capitalised interest up ~7x. $5.3M in Q2 and $10.1M in 1H, versus $0.7M / $1.3M a year earlier. Same mechanism, applied to interest.
4. Bad debt is deteriorating sharply. Provision for estimated credit losses $29.8M in 1H 2026 vs $0.4M in 1H 2025; allowance $12.6M → $22.9M; write-offs $19.5M. Accounts receivable also consumed $58.6M of cash on "unfavorable timing differences in the collection of accounts receivable." Consumer stress or billing-integration friction — the filing does not say which.
5. RDOF milestone shortfall — new regulatory exposure. "Uniti met the 40% completion on or before December 31, 2025 in a number of states but permitting delays in other areas led to a shortfall. Uniti notified the FCC and Universal Service Administrative Company on January 15, 2026". RDOF is $522.8M over ten years / $52.3M a year for ~192,000 locations. The filing discloses no quantified penalty and no reserve. This is the first named non-compliance event in the file and the prior dossier did not have it.
6. Asset encumbrance as a structural flag. 15.5% of assets and 45% of operating income now sit in non-guaranteeing unrestricted subsidiaries (Lens 4 / delta #7). Nothing improper — it is fully disclosed in Note 14 — but a reader modelling recovery value for the common must now model two balance sheets, not one.

Regulatory findings (re-verified): regulatory/regulatory-findings.md was regenerated 2026-08-10 covering 2021-08-10 → 2026-08-10. SEC Litigation Releases: none. SEC AAERs: none.. 10-Q Part II Item 1 incorporates Note 13 and discloses no material proceeding; Item 1A states there have been no material changes to the risk factors from the FY2025 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes.. Disclosure controls concluded effective; no changes to ICFR in the quarter. The non-SEC agency sweep the findings file calls for could not be completed this run — the web-search budget was exhausted before it ran. That is an unchecked box, not a clean result.

Net: still no fraud signal and no enforcement. The forensic risk remains disclosure interpretation — and Q2 added three separate legitimate accounting choices (capitalised labour, capitalised interest, penalty/reserve revenue) that each push reported margin up while cash goes the other way. None is a red flag alone. Together they mean Adjusted EBITDA is a worse proxy for cash generation this quarter than last, which matters because Adjusted EBITDA is the number the entire bull case and every debt covenant runs on.

Forward Projection — refreshed

Model status, stated before any number is cited. model.xlsx was rebuilt this run (our model) and read back (our model). It returned "No computed values — the workbook has 51 formulas and no cached results." It also flags: only 2 quarters of financials (LTM figures partial), revenue growth seeded at a default 8% for want of 8 quarters, opening balance sheet only partly sourced (8/10 lines), and no share price set. Nothing below is taken from the workbook. Every figure is hand arithmetic with the calculation shown. The workbook is now seeded for the next refresh, which is its only present value.

Why EPS is the wrong scoreboard — and the answer to the question anyway. Management guides FY2026 to a net loss of $(530)–(480)M. Below EBITDA sit ~$1.22B of D&A (1H $595.7M) and $805M of guided net interest — ~$2.0B of fixed charges against $1.45–1.50B of EBITDA. No plausible three-year path produces positive GAAP EPS. Presented anyway, per the battery:

[All ``. FY2026 = guidance range net loss ($530/505/480M) plus preferred dividends ~$63M (Q2 $15.7M × 4), ÷ 253M weighted-average basic shares. FY2027–28 built as: Adj EBITDA − D&A − net interest − preferred, ÷ 253M. Base: EBITDA 1.55B/1.75B, D&A 1.32B/1.40B (rising with the capex programme), interest 0.82B/0.80B. Bear: EBITDA 1.42B/1.45B, D&A 1.35B/1.45B, interest 0.85B/0.90B. Bull: EBITDA 1.68B/2.00B, D&A 1.30B/1.40B, interest 0.79B/0.75B. Preferred escalates 63→66→70M. Shares held flat — no buyback capacity, modest equity-comp dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. ignored.]

The number that actually decides the equity: net leverage.

[``. Starting net leverage = $10,072M ÷ ~$1.44B clean run-rate = 7.0x; on the FY2026 guide midpoint of $1.475B it is 6.8x.]

This is a full turn-and-a-half worse than the prior dossier, which had 5.5x → 4.5x (base) and <4.0x (bull). Even the bull case no longer reaches 5x by 2028. The change is not primarily operational deterioration — it is the correction of an inflated EBITDA denominator (Q1 annualised at $1.77B vs a clean ~$1.44B).

Free cash flow. FY2026: EBITDA $1.475B − net interest $805M − capex ~$1.45B (Kinetic $1.27B + Fiber Infrastructure $0.14B + corporate/shared) ≈ −$780M, before working capital. 1H actual levered FCF was −$459.8M, so the full year is consistent. Levered FCF does not turn positive on any of the three paths inside the horizon; the bull case merely narrows it. The build is funded by ABS issuance and asset sales, not operations — which is exactly what management said.

Scoreable forecast (logged conceptually — our model create intentionally NOT run, per the unattended-run rules). The prior dossier's binary was "FY2026 Adjusted EBITDA ≥ $1.45B, p≈0.65." With 1H at $798.7M and the guide raised to $1.45–1.50B, that binary is now nearly settled — 2H needs only $651.3M, i.e. $326M/quarter versus $357.1M just printed, p≈0.88. The binary worth carrying forward instead is the one that tests the thesis rather than the guide: "UNIT Q4 2026 total service revenue ≥ Q4 2025 pro-forma service revenue, p≈0.20" — five straight quarters of −8% to −9% service-revenue decline, with ARPU pressure acknowledged and consumer fibre unable to outrun DSL erosion, make stabilisation inside two quarters unlikely.

Bull vs Bear — refreshed

Bull case (updated). The asset is compounding even as the P&L shrinks, and the financing has been repaired further than the market has priced. Q2 set records on every operational metric that matters — 141k passings, 38k net adds, 603k fibre subs (+25%), $2.2M MRR of new wholesale bookings (+30% over the prior record) — and the demand mix is improving, not just growing: >50% of new bookings are Waves/lit rather than dark fibre, at 37% blended anchor lease-up cash yields on largely already-built plant. That is the AI-interconnect thesis showing up in the bookings line rather than the narrative. On the liability side the story is genuinely better than the last dossier believed: blended debt yield 6.5% versus 12.5% three years ago, ~$3B of securitisation raised at 5.2–6.2% for the senior tranches, no 2028 cliff (the $2.2B 8.25% notes mature 2031), covenant compliance, and ~$1.45B of liquidity. Management has identified $500M–$1B of non-core assets it says produce "minimal to no cash flow" — deleveraging with no EBITDA cost, the cleanest lever available. Kinetic faces cable in under 60% of its fibre footprint versus mid-80s to low-90s for peers. Elliott's ~25% stake keeps a strategic-sale tail alive. At ~$2.28B of market cap on ~$8.6B of net PP&E, the equity is a cheap option on the fibre being worth more than the debt.

Bear case (updated, and the burden of proof has shifted).

1. The thesis-critical number is going backwards. Recurring service revenue −9% YoY; Adjusted EBITDA −10% pro forma; the one quarter of growth that validated the merger was a one-off IRU. Deleveraging-by-EBITDA-growth requires EBITDA growth.
2. Real leverage is ~7.0x, not 5.5x, and Q2 operating cash flow ($78.5M) did not cover Q2 interest ($195.6M). Book equity is $161.5M — two more quarters at the Q2 loss rate and reported equity is negative.
3. The collateral is leaving the restricted group. 15.5% of assets and 45% of operating income now sit in non-guaranteeing ABS subsidiaries, and the July deal moved five more states of Kinetic fibre in. Each securitisation lowers the blended coupon and thins what stands behind the senior notes — and behind them, the common.
4. Capex intensity increased rather than relenting — $450M in Q2 against $357.1M of EBITDA, with FY guidance raised another $100M. The "pull-forward of 2027 spending" framing is unfalsifiable until 2027.
5. Pricing power is confirmed absent. ARPU is falling on cable/telco promotions inside Uniti's own footprint, with recovery promised for Q4 2026 and only "low single digits" thereafter.

Pre-mortem (18 months out, thesis broke). It is early 2028. The Q4 2026 hyperscaler dark-fibre deals slipped to 2027 and then shrank; without them, 2027 revenue printed down again and Adjusted EBITDA finished at ~$1.42B rather than the $1.55B the base case needed. Kinetic kept adding subscribers — penetration reached 33% — but ARPU never recovered past low-single-digit growth, and DSL ran off faster than fibre replaced it, so the segment stayed in aggregate decline. Uniti Solutions shrank at 16% a year and became immaterial. Capex did not taper in 2027 because the pull-forward turned out to be a level-shift; levered FCF stayed near −$700M and was funded with a third and fourth Kinetic ABS, moving the remaining unencumbered fibre into the unrestricted box. Net leverage sat at 7.5x. The $500M–$1B non-core monetisation produced $250M at disappointing prices because the assets genuinely produced no cash flow and buyers priced them accordingly. Then the $886.5M 2027 maturity and the 2028 $1.12B arrived; refinancing was available but at spreads that put net interest through $900M. Goodwill on Fiber Infrastructure — already carrying $515.5M of prior impairment — took a fresh write-down. The common, a $2.3B wafer on $11.5B of debt with the best collateral pledged elsewhere, traded to option value. Elliott, unable to exit cleanly, pushed for a sale of Fiber Infrastructure — the only segment that grew — and the integrated story that justified the merger came apart.

Are multiples too high? This is the answer that changed. Previously: "No — UNIT is optically cheap." Now: at ~8.8x EV/forward-Adj-EBITDA (~9.0x on the clean run-rate) for a business with declining recurring revenue, declining EBITDA, ~7x leverage and deeply negative levered FCF, the multiple is not cheap — it is roughly full. The share price fell 19% and the multiple rose, because the prior denominator was wrong. The margin of safety the last dossier saw was largely arithmetic.

Contrarian view (what the market may still be refusing to see). The bear case above is the consensus case now — six analysts at Hold, JPMorgan cutting, the stock down 19%. What may be underpriced is the Fiber Infrastructure segment in isolation: growing 10%, 52% contribution margin, record bookings pivoting to lit capacity, 37% anchor lease-up cash yields, and named as the demand surface for agentic-AI inference at the edge. It is ~22% of revenue attached to ~78% of consolidated dead weight. If the $500M–$1B monetisation becomes a larger break-up — or if Elliott forces a separation — the market is currently valuing that asset inside a melting ILEC. The bull case for UNIT may no longer be "the deleveraging works"; it may be "the sum of the parts gets tested." That is a different, and more Elliott-shaped, thesis than the one in the prior dossier.

Devil's Advocate (short seller) — refreshed

Dismantling the updated bull case:

• "Record operational metrics" is the tell, not the proof. When a company leads with passings, net adds, bookings MRR and cash yields, and does not lead with revenue or EBITDA, it is because revenue and EBITDA went the wrong way. Every record cited is an input metric. The output metrics — service revenue, Adjusted EBITDA, operating income, free cash flow, book equity — all deteriorated. Records on inputs while outputs fall is the signature of a business buying growth at a price it has not yet proven it can earn back.
• The 37% anchor lease-up yield is the strongest bull number and it does not scale. It is high because 80% of that hyperscaler business sells existing, already-built infrastructure. That is a stock of underutilised assets, not a flow. The moment lease-up requires new build, the yield converges toward the cost of the build.
• Structurally: the business does not cover its interest in cash. Q2 OCF $78.5M vs $195.6M interest. The prior dossier said the business doesn't earn its interest at the operating line; that has now reached the cash line, and it happened in the quarter after the refinancings that were supposed to fix it.
• The financing "improvement" is partly a mortgage. Yes, the blended yield fell to 6.5%. It fell because Uniti moved its best consumer fibre — ten states now — into bankruptcy-remote boxes and borrowed against it at 5.2–6.2%. Cheaper debt secured on better collateral is not the same as a better balance sheet. The unsecured 8.625% notes still exist; what changed is how much stands in front of them. The equity is the last claim on whatever is left outside the ABS boxes.
• Most dangerous competitor bulls underestimate — revised. The prior dossier named T-Mobile/Verizon FWA. Q2 says the damage is being done by something more mundane and more immediate: cable and telco promotional pricing, which Uniti is matching, in its own incumbent footprint, right now. FWA is the structural ceiling; the promotional war is the present bleed. Harrobin's "less than 60% overlap with cable" is a real advantage, but it also means cable is in 60% of the footprint, and that is where the ARPU is going.
• What must hold for today's ~$9.39 price: service revenue must stop falling; the Q4 hyperscaler deals must land and not slip; capex must actually taper in 2027; the $500M–$1B non-core sale must fetch real money for assets management has publicly described as producing no cash flow; ARPU must inflect in Q4; the RDOF shortfall must stay unpenalised; and the 2027–2028 maturities must refinance near current spreads. Seven independent things, up from five in the prior dossier.
• The $500M–$1B monetisation contains its own contradiction. Bullington's reassurance is that these assets have "negligible effect on our adjusted EBITDA" because they are "underutilized today and currently produce minimal to no cash flow." Buyers will apply exactly that logic to price. You cannot simultaneously tell equity holders the assets are worth $1B and tell them the assets produce no cash flow — one of those statements is doing promotional work.
• Valuation if growth disappoints 20–30%: at FY2028 EBITDA of ~$1.3B against ~$11.5B of net debt, leverage is ~8.8x, refinancing is punitive, and a $2.28B equity is impaired 60–80% — worse than the prior dossier's 40–60%, because the starting leverage is a full turn higher and the best collateral is already pledged.
• Single scenario that permanently impairs: not a refinancing shock this time — the 2031 correction removed that cliff. It is serial securitisation. Each ABS lowers the coupon and buys another year of build; each also moves fibre out of the restricted group. Four more deals and the common is a residual claim on an ILEC in run-off plus whatever equity remains in the ABS structures. Plausibility: moderate-to-high, because it is not a risk — it is the stated, disclosed, currently-executing plan.

---

Lenses carried by reference (unchanged)

• Lens 1 · Company Overview — unchanged, see the previous dossier. (~240,000 route miles, 47 states, three segments, de-REIT'd, no common dividend.)
• Lens 2 · Supply Chain — unchanged, see the previous dossier. The Q2 filing reinforces rather than revises it: outside contractors up (Kinetic sales revenue +154% is contractor equipment sales), permitting named as the binding constraint on both the RDOF shortfall and possible Q4 deal slippage.
• Lens 14 · Management Questions — unchanged, see the previous dossier. Three to promote to the top given Q2: (a) what is the like-for-like Adjusted EBITDA margin adjusted for the increase in capitalised internal labour?; (b) how much fibre remains in the restricted group, and is there a floor below which you will not securitise?; (c) what is the RDOF shortfall in locations and dollars, and what support is at risk?

---