EXCL The Loss Shrank 61% This Quarter - Normalized Profit Didn't Move At All

The Loss Narrowed - The Underlying Business Didn't Grow

PT XLSmart Telecom Sejahtera Tbk (EXCL) - the merged entity formerly known as PT XL Axiata Tbk - reported Q2 2026 (the three months ended June 30, 2026) revenue of Rp12,173 billion, up 3.0% from Q1 2026's Rp11,819 billion, and a net loss attributable to owners of Rp277 billion, down 61.3% from Q1's Rp717 billion loss. Read at face value, that looks like a company turning a corner. It isn't quite that.

The company's own normalized profit after tax measure - which strips out merger-integration costs and accelerated depreciation - was Rp1,354 billion this quarter, essentially flat (-1%) from Q1's Rp1,375 billion. The reported loss shrank almost entirely because the accelerated depreciation charge tied to the merger fell from Rp2,069 billion in Q1 to Rp1,597 billion in Q2, not because the underlying mobile business generated more profit. EBITDA margin held at 46% for a second straight quarter - genuinely a sign of stability - but stability isn't the same story as the "resilient performance with improving earnings quality" framing the company's own investor presentation leads with. None of this shows up cleanly in a year-over-year comparison, either: Q2 2025, the comparative quarter, is an even messier YoY base than Q1 2025 was, since XL Axiata's merger with Smartfren and Smart Telecom became legally effective April 16, 2025 - squarely inside that quarter - so Q2 2025's results are neither cleanly pre-merger nor cleanly post-merger. They blend roughly six weeks of standalone XL Axiata with about ten weeks of a partially-consolidated combined entity, plus a one-off Rp802 billion merger-related asset impairment that alone pushed that quarter to an operating loss (before finance costs) of roughly Rp834 billion, on the way to a reported net loss of about Rp1,608 billion. None of that is informative about how the fully-merged business is actually performing today, which is why every comparison below leans on the sequential Q1-to-Q2 2026 view - the first quarter pair in this company's history where both sides are genuinely the same combined entity - while still disclosing the YoY figures for completeness.

The Prescription

XLSmart should keep defending ARPU^» discipline, but not declare victory on it: blended mobile ARPU actually dipped 0.8% quarter-over-quarter to Rp46,900 after Q1's Rp47,300, and postpaid ARPU - the higher-value segment - fell a sharper 9.2%, from Rp89,800 to Rp81,500. The subscriber base held flat at 69.4 million (0% QoQ, still down 16% from a year ago as the post-merger "market repair" of low-value SIMs continues), so this isn't a volume story masking a price story; it's a genuine early wobble in the one strategy this project's Q1 2026 post flagged as clearly working. One quarter of ARPU softness isn't a trend yet, but it's the first data point worth watching next quarter.

What it should stop doing: adding committed forward obligations faster than the deleveraging can absorb them. The gearing ratio^» - total debt including finance leases divided by annualized quarterly EBITDA - actually improved this quarter, from 2.84x in Q1 to 2.66x, the first sequential decline since the merger closed. That's real progress. But in the same quarter, the company's disclosed network-expansion purchase commitment grew from USD1.74 billion to USD1.94 billion, FY26 capitalized-capex guidance was raised from roughly Rp15 trillion to roughly Rp20 trillion, and - as a subsequent event - XLSmart won a 10-year, 700MHz/2600MHz 5G spectrum license carrying a Rp1,292 billion upfront fee plus a Rp993 billion annual usage fee through 2036 (see Beyond the Usual), funded in part by fresh bank loans drawn in the following weeks. Free cash flow fell 68.8% quarter-over-quarter as a result - capitalized capex more than doubled sequentially (Rp2,242 billion to roughly Rp5,551 billion in cash terms) while operating cash flow barely moved. A single quarter of improving leverage sitting next to a rapidly growing forward commitment schedule is not a contradiction yet, but FY26 guidance still says nothing about where gearing is supposed to land once all of this is actually paid for - the same gap flagged last quarter, now with a bigger number attached to it.

Key Financial Metrics

Q2 2026 (three months ended Jun 30, 2026) vs Q1 2026 (sequential) and Q2 2025 (YoY, not comparable - see above)

The company's interim financial statements only disclose six-month cumulative figures for H1 2026 and H1 2025 - there is no separately filed three-month statement. Every Q2 2026 and Q2 2025 figure below is derived as H1 minus the already-published Q1 figure for that year (H1 2026's Rp23,992 billion revenue minus Q1 2026's Rp11,819 billion, for example), reconciled against the filing's own H1 subtotals before use.

FX: Rp17,856 = US$1 (Jun 30, 2026), the period-end rate disclosed in the company's own interim financial statements, used only to convert the USD columns.

*YoY compares Q2 2026 against Q2 2025 - a quarter that straddles the merger's April 16, 2025 effective date and carries a one-off Rp802 billion asset impairment, so it is neither a clean pre-merger nor a clean post-merger baseline. Treat every YoY figure here as directional at best; the QoQ column is the fair comparison.

Why each moved: Revenue grew a modest 3.0% sequentially, in line with a normal (non-festive) quarter following Q1's Ramadan-boosted base - Q2 has no comparable seasonal tailwind. Operating income jumped almost 5x sequentially (Rp108 billion to Rp528 billion) because the accelerated depreciation embedded in operating expenses eased from Rp2,069 billion in Q1 to Rp1,597 billion in Q2, not because revenue or the underlying cost base improved by a similar magnitude. EBITDA margin held at 46% for the second straight quarter, matching the low end of the company's newly-updated FY26 guidance range (see The Prescription). Free cash flow fell sharply because capitalized capex, on a cash-paid basis, roughly doubled sequentially (from Rp2,242 billion to an estimated Rp5,551 billion) while operating cash flow was essentially flat - the clearest line item connecting this quarter's numbers to the capex-guidance increase and spectrum win discussed below. Total cash fell 25.4% for the same reason: more of this quarter's cash went into fixed-asset purchases and loan repayments than came in from operations.

Key Operational Metrics

• Mobile subscribers: 69.4 million, flat (0%) quarter-over-quarter, down 16% from 82.6 million a year ago as post-merger "market repair" (removing low-value/inactive SIMs) continues. Unlike the P&L, this YoY comparison is genuinely apples-to-apples - Q2 2025's subscriber count was already a post-merger, point-in-time figure, since the merger had closed within that quarter.
• Blended mobile ARPU^»: Rp46,900, down 0.8% quarter-over-quarter from Rp47,300 - the first sequential dip since the merger, driven mainly by postpaid ARPU falling 9.2% (Rp89,800 to Rp81,500); prepaid ARPU was roughly flat (Rp46,100 to Rp45,800). Still up sharply from a year ago (Rp35,500 blended, 2Q25).
• Mobile data traffic: 4,067 petabytes, up 5% quarter-over-quarter and 7% year-over-year - a genuine acceleration, not a seasonal artifact, since Q2 lacks Q1's Ramadan-driven traffic surge.
• Fixed broadband subscribers: 960,000, up 2% quarter-over-quarter from Q1's 940,000 - a small recovery in the one operational line the Q1 2026 post flagged as not yet showing post-merger improvement, though still below the year-ago 980,000.
• Network integration: 92% of sites now integrated onto the combined network (up from 77% of targeted tower dismantling completed as of Q1), with total BTS count reaching 265,442 (+26% YoY, a much slower +4.6% QoQ pace than Q1's +12% as the network shifts from heavy build-out to what management calls an "optimization phase"). 5G footprint expanded to 52 cities (from 43 at Q1-end), with ~15,000 5G BTS delivering 23% national population coverage and contributing 27% of traffic growth.
• Seasonality: Q2 has no equivalent to Q1's Ramadan/Lebaran traffic and top-up surge, so the more modest sequential gains in revenue and traffic this quarter should be read against a naturally elevated Q1 base, not as a slowdown.

Segment Comparison

XLSmart reports two segments: GSM mobile and telecommunications network services (99.1% of Q2 revenue) and managed service and information technology services (1.1%, serving enterprise customers). The Group operates in a single geographic market, so no geographic segmentation is reported.

The core mobile business again carries essentially the entire reported loss, consistent with Q1. The smaller managed-service segment's own loss actually widened from roughly Rp2 billion (Q1) to roughly Rp8 billion (Q2) even as its revenue grew 9.8% sequentially (Rp119 billion to Rp131 billion) - a reversal of the prior quarter's slight revenue decline, but paired with worse segment economics. At just over 1% of consolidated revenue this doesn't move the overall picture, but it's worth tracking whether the segment can grow revenue and improve its result at the same time.

Beyond the Usual

A 10-Year Spectrum License Arrived Right After Quarter-End, With a Rp2.3 Trillion First-Year Price Tag

As a subsequent event, Indonesia's Ministry of Communication and Digital Affairs awarded XLSmart the 700MHz and 2600MHz 5G spectrum bands on July 21, 2026, through a competitive auction. The license runs from August 5, 2026 to August 4, 2036 and requires an upfront fee of Rp1,292 billion plus an annual usage fee of Rp993 billion. In the weeks that followed, the company drew Rp1,000 billion from a Bank DBS Indonesia facility (granted May 12, 2026) and Rp200 billion from a Bank Permata facility (granted May 22, 2026) - the first visible funding for this new obligation, and a direct match to the presentation's own framing of "increased investment following the spectrum acquisition."

The Network-Expansion Commitment Keeps Growing Faster Than Any Single Year's Capex Guidance

The Group's disclosed network-expansion purchase commitment grew from USD1,738,794,197 (equivalent to Rp29,547,330 million) at Q1 2026 to USD1,942,402,754 (equivalent to Rp34,683,544 million) at Q2 2026 - an increase of roughly USD204 million in a single quarter, even before the new 5G spectrum obligation above is added on top. This is a real, contractually committed forward liability that keeps growing faster than any single year's guided capex spend, in a quarter where the gearing ratio (see The Prescription) improved for the first time since the merger. Whether that improvement continues once these commitments start converting into actual cash outflows is the open question for the next few quarters.

The Company's Own Ownership Note Puts Two Blocs at Near-Parity, Neither in Control

XLSmart's own general disclosures name four affiliated entities - PT Bali Media Telekomunikasi (24.57%), PT Global Nusa Data (4.66%), PT Wahana Inti Nusantara (2.85%), and PT Gerbangmas Tunggal Sejahtera (2.61%) - collectively as the "Sinarmas Group," which together hold 34.69% of shares outstanding. That figure lands almost exactly level with Axiata Investments (Indonesia)'s own 34.69% stake, meaning the two largest shareholder blocs are effectively tied, with the remaining roughly 30.6% held by minority directors and the public (each individually under 5%) and no single party in control. The company states plainly that these are entities with significant influence, not control - a structure worth tracking for how capital-allocation and governance decisions get made when neither of the two largest holders can outvote the other alone.

The Rupiah Actually Strengthened Slightly After Quarter-End This Time

As a subsequent event, the company discloses the IDR/USD rate moved from Rp17,856 (June 30, 2026) to Rp17,795 by August 11, 2026 (the date the interim financial statements were signed) - a small strengthening of the Rupiah, the opposite direction from the weakening flagged in the Q1 2026 post. Restating the Group's foreign-currency monetary position at the later rate would increase unrealised foreign exchange profit by a modest Rp1,475 million - immaterial next to Q1's Rp18,570 million swing, but consistent with the same underlying exposure (mostly US-dollar-denominated trade payables against a smaller base of US-dollar assets) that makes this company's earnings sensitive to Rupiah movements either direction.

The Group's net FX-exposed monetary liability position, measured in Rupiah-equivalent terms, fell from Rp714,372 million at December 31, 2025 to Rp433,737 million at June 30, 2026 - a reduction of nearly 40% in the underlying balance-sheet exposure itself, separate from the exchange-rate movement above.

Target Valuation Range

Bottom line: EV/EBITDA scenario math implies a roughly Rp1,172-Rp4,260 fair-value range (bear-to-bull) against a Rp2,370 actual close - sitting between the bear and base cases, closer to base, and meaningfully cheaper in absolute multiple terms than Q1's already-compressed range, even though this quarter's own gearing improved and EBITDA margin held steady.

This is now the third consecutive quarter with a fully post-merger, normalized profitability base, still too short a track record for a full multi-year DCF. What follows is the same labeled EV/EBITDA sanity check as the prior two posts, extended with this quarter's data point.

Market-cap and enterprise-value buildup (18.20 billion shares outstanding, unchanged since the merger):

Peer-multiple comparison (revenue and EBITDA annualized ×4 from the quarter, extending the trailing series the last two posts started):

The multiple has now compressed for three straight data points - 5.89x to 5.17x to 4.47x - even as normalized EBITDA kept growing slightly each quarter, meaning the de-rating is entirely a price story (Rp3,750 to Rp2,950 to Rp2,370) rather than a deteriorating business. P/E remains unusable since the company is still reporting a net loss.

Peer check against Indosat: using the same ex-lease, half-year-annualized convention Indosat's own Q2 2026 post uses for its own multiple (total debt excluding lease liabilities, EBITDA annualized ×2 from H1), XLSmart's H1 2026 reported EBITDA of Rp10,971 billion annualizes to Rp21,942 billion, and an EV built the same way (market cap Rp43,134 billion, plus ex-lease debt of Rp20,868 billion, less cash) comes to roughly Rp61,985 billion - an EV/EBITDA of approximately 2.83x, against Indosat's own disclosed ~2.00x. XLSmart still trades at a premium to Indosat's multiple on this basis. A precise premium-versus-last-quarter comparison isn't reliable here, since Indosat's own Q1 2026 post used a different annualization window than its Q2 2026 post - the more reliable read is that XLSmart's own absolute multiple keeps compressing every quarter regardless of which peer or convention it's measured against.

DCF-lite scenarios (a single EV/EBITDA multiple applied to annualized normalized EBITDA, not a full discounted cash flow):

Reading the table: the Rp2,370 actual close sits between the bear and base cases, closer to base - the market appears to already be pricing in meaningful skepticism about whether this quarter's flat normalized profit and growing forward commitments will resolve favorably, without yet assuming the bear case's further deterioration. The spectrum obligation and network-expansion commitment flagged in Beyond the Usual are the clearest catalysts that could push this toward either end of the range over the next few quarters, depending on whether the easing gearing trend holds once real cash starts going out the door for both.