Date: 2026-02-22 Period: Q3 2025 (quarter ending September 30, 2025) Task: Earnings review Position: No position (watchlist)
briefs/ARX_earnings-review_2026-02-21/companies/ARX.mdearnings/ARX/_ROLLING.mdLet me be upfront — ARX sits outside my typical hunting ground of cloud/SaaS/infrastructure. This is an insurance exchange. But here's the thing... the platform dynamics are more interesting than the sector label suggests.
What Accelerant actually is: A specialty insurance exchange that connects specialty insurance "members" (small E&S carriers) with risk capital (Lloyd's syndicates, reinsurers, Hadron — their own capital). Members write policies; Accelerant provides the capital, data infrastructure, and analytics backbone. Revenue comes from two sources: (1) taking a cut of Exchange Written Premium (~8% take rate → Exchange Services), and (2) underwriting profit from their own risk capital (Hadron segment).
The crowdsourced intelligence angle is REAL. This is
my framework center mass. They're collecting 57,000 risk attributes
across 265 members. Each member who joins makes the data network smarter
for everyone — better loss prediction, better pricing, tighter risk
selection. That's a genuine data flywheel I haven't seen discussed
enough. "23,000 → 57,000 attributes" is the kind of metric that deserves
!!.
Building block or process tool? Members run their entire specialty insurance operation ON Accelerant's platform (Hadron as the risk spine). This is closer to a building block — their business is embedded in the exchange. Switching costs are high: you lose your data history, your capital relationships, your pricing benchmarks. That's what drives 135% NRR.
Self-serve GTM? Mostly yes — Accelerant doesn't sell through big SI relationships. Members come to the exchange organically. New third-party insurers (Lloyd's, Ozark, Incline, MS Transverse in Q3 alone) are joining because the data quality makes their capital deployable at lower risk. Network effects working.
| Metric | Q3 2024 | Q4 2024 | Q1 2025 | Q2 2025 | Q3 2025 | Trend |
|---|---|---|---|---|---|---|
| EWP ($m) | 796 | ~820 | ~862 | 912 | 1,043 | ^^ |
| EWP YoY% | ~40% | ~40% | ~45% | ~35% | 31% | ↓ |
| Rev (Exchange Svcs, $m) | ~63 | ~71 | ~74 | ~80 | 85 | ^^ |
| Take Rate | 7.1% | 7.3% | 7.5% | 7.8% | 8.0% | ^^ !! |
| Adj EBITDA ($m) | ~24 | ~28 | ~35 | ~50 | 41* | — |
| Adj EBITDA Margin | ~24% | ~24% | ~25% | ~35% | ~25%† | stable |
| NRR | 151% | 148% | 140% | 138% | 135% | ↓ |
| Members | 204 | 215 | 235 | 250 | 265 | ^^ |
| Gross Loss Ratio | ~57% | ~54% | ~52% | ~51% | 50.1% | ^^ !! |
| Net Loss Ratio | ~72% | ~70% | ~65% | ~63% | 61.9% | ^^ |
*41mincludes 39m one-time investment gain — underlying ~$25m EBITDA at ~24.7% margin †Underlying margin (ex-gains)
Revenue growth: 74% YoY [Non-GAAP basis — includes investment income and underwriting profit, not pure SaaS revenue]
Here's the thing — the 74% headline is real but I want to be precise about what's happening. Exchange Services revenue (the take-rate income) was $85m, +34% YoY at 8.0% take rate. That's the cleanest signal. The broader revenue figure is inflated by underwriting operations, investment income, and a $39m non-recurring investment gain.
The clean scorecard:
!!^^ — this is the
operating leverage story!!NRR 135% — flag but not alarm. Down from 151% a year ago. The decline makes sense: 80% of new EWP comes from existing members expanding their books. As the member base matures, individual member growth rates naturally normalize. 135% is still exceptional — it just means the law of large numbers is asserting itself.
Pipeline: $3B annualized, $1.8B under contract. That's almost 2x current run-rate EWP in the pipeline. The third-party insurer expansion is accelerating — 4 new capital providers in Q3 alone. This is the strategic scale unlock.
Management said clearly: current third-party mix is 32% of EWP. Target is 67% in 3-5 years. This is the platform flywheel going into overdrive.
Why does this matter more than anything else on the call?
Third-party capital is essentially fee-for-service with zero balance sheet risk. Accelerant facilitates the match (their data predicts losses), takes the take rate (~8%), and passes the risk to Lloyd's/reinsurers. Pure platform economics at scale.
Every dollar of third-party EWP added is at lower margin but zero capital drag. FY26 EBITDA guidance ($269M, 5.4% of EWP) looks weak versus Q3's underlying ~25% — but that's because Exchange Services is the high-margin segment and the mix is shifting toward third-party volume.
Hadron (their own risk capital) is the proof of concept. They've already demonstrated they can price this risk correctly (50% gross loss ratio). Now they're licensing that capability to external capital at scale. That's the platform model working.
NUANCE: The 5.4% EBITDA margin on FY26 EWP (5B + ×5.4270M) is NOT the right way to read profitability. The right read is: Exchange Services at 47% EBITDA margin, scaled to ~$400M revenue at 8% take rate on 5BEWP.That′s 188M from Exchange Services alone at high margin. Plus Hadron underwriting profit on top. Management is being conservative on the mix-shift math.
Data attributes are the real moat. 57,000 risk attributes across 265 members. Each new member makes the model better for ALL members. This creates a compounding competitive advantage that makes it increasingly hard for a new entrant to replicate. It's not just scale — it's the quality of the prediction engine.
The member discipline story. 15 members asked to leave since 2018. That's an anti-NRR move (short-term revenue loss) in service of pool quality. That level of underwriting discipline is unusual and directly feeds the virtuous cycle (better data → lower losses → more attractive to capital → better terms for members → more members). Long-term thinking embedded in the operating model.
Operating leverage signal: take rate 7.1% → 8.0%. This is accelerating and not fully appreciated. As Accelerant's data advantage widens (better risk prediction), they can justify a higher take for the certainty/quality they deliver. This is pricing power from analytical advantage.
| Metric | Q4 2025 Guide | FY 2026 Guide |
|---|---|---|
| EWP | $1.06-1.10B | $5.0B+ |
| Adj EBITDA | $57-62M | $269M |
| EBITDA Margin (on EWP) | ~5.5% | ~5.4% |
FY 2026 EWP guidance of 5B + implies 613.1B annualized). That is a step-change acceleration.
The margin compression in Q4 vs Q3 is attributable to: mix shift to lower-margin third-party, absence of Q3 investment gains, and infrastructure investment ahead of the $5B EWP ramp. I accept management's framing here — the underlying Exchange Services margin expansion (take rate trajectory) tells the right story.
Management has 3 quarters of guidance history (limited). The Q4 guide of $1.06-1.1B EWP is ~1.7-5.4% above Q3's $1.04B — slight deceleration, consistent with seasonality in specialty insurance. Not alarming.
NRR declining 151% → 135% — the trend is downward. I need to see stabilization at 120%+ to stay comfortable. If it dips below 115%, the existing member expansion story is weakening.
7 quarters public — we have almost no cycle data. Specialty insurance looks great in benign loss years. What happens when CAT events spike? The loss ratio resilience (50% gross) is good but untested at scale through a catastrophe year.
Hadron capital concentration — 64% of EWP goes through their own risk capital. As they scale to $5B and shift to 67% third-party, Hadron's share drops. Good for margin quality, but Hadron is also where the highest margins live. Watch the mix math carefully.
Take rate ceiling? The 8% take rate is already high for an exchange model. Is 9%+ achievable, or does this plateau? Management hasn't guided explicitly on this.
This is unusual for my framework because ARX is not SaaS/cloud. But applying the framework:
!! — better than most SaaS
names I track!!My tier: Tier 2/3 hybrid — Tier 2 on platform quality, Tier 3 on reported financial metrics due to insurance accounting complexity and 5.4% EBITDA/EWP optics
The 5.4% EBITDA margin looks terrible through a SaaS lens. But Exchange Services at 47% EBITDA margin on a rapidly expanding take rate — that's the right comparison.
No prior muji writings on ARX exist. First look at this company.
Closest analogy in corpus: marketplace platforms with data network effects (similar to discussions of Cloudflare's network, CrowdStrike's crowdsourced threat intelligence). The "crowdsourced intelligence" angle maps directly — Accelerant is doing for specialty insurance what CrowdStrike did for endpoint security: pooling anonymized data from all participants to make predictions that no individual could make alone.
Watchlist — tracking, not yet allocated. ARX has genuine platform characteristics I respect: data flywheel with 57k attributes, 135% NRR, take rate expansion, member discipline that prioritizes pool quality over short-term revenue. The third-party pivot is the right strategic move and could deliver a step-function in platform economics.
What's keeping me from pulling the trigger today: (1) only 7 quarters public — no cycle data, (2) the insurance accounting makes my financial framework harder to apply cleanly, (3) I need to see the NRR stabilize at 130%+, and (4) I'd want to understand how CAT exposure is managed before sizing a position.
If NRR stabilizes and Q4/Q1 FY26 EWP confirms the acceleration toward $5B+, this could move to a 3-5% starter position. The platform dynamics are real. The data moat is real. The market probably isn't pricing the take rate expansion correctly.
DING DING DING on the data flywheel story. Just need more runway to validate.
-muji No position in ARX