This is not an arm's-length merger between two independent companies. Both sit inside the same private equity portfolio. What is being reported is a re-cut of assets New Mountain Capital already owns:

• SmarterDx is pulled out of Smarter Technologies — the RCM platform NMC assembled in May 2025 from SmarterDx, Access Healthcare and Thoughtful.ai, with $800M+ combined revenue, 200+ clients and 60+ hospitals and health systems.
• It is folded into Datavant.
• The move is financed by a ~$2.5B debt recapitalization of Datavant's existing facility, run by JPMorgan.

The distinction matters for interpretation. No price was negotiated between opposing parties, so the transaction carries no external validation of value. It is a sponsor moving a growth asset onto the balance sheet of a cash-generative one.

With the Thoreau exit gone, New Mountain holds five mature healthtech assets, limited partners expecting distributions, and no liquidity path. The recap is the substitute for the sale: lever the highest-EBITDA asset in the portfolio and return capital without selling it.

The SmarterDx transfer is what makes that levered entity financeable and, eventually, exitable — it attaches a growth story to a mature compounder. The strategic logic examined in section 4 is real and defensible. But the sequencing indicates which consideration came first.

Leverage. $2.5B against ~$500M of EBITDA is approximately 5.0x — serviceable on its own. But SmarterDx is EBITDA-dilutive, so pro-forma leverage is worse than 5.0x, not better. The structure levers a cash-generative asset and then adds a cash-consuming one to it. That works only if SmarterDx's margin scales quickly, or if the combined narrative supports a higher exit multiple. The bet is visibly on the second.

Implied value. Datavant at 15–20x EBITDA, consistent with health-data infrastructure comparables, is $7.5–10B. SmarterDx at 8–12x revenue is $1.2–1.8B, anchored on Waystar's $1.25B enterprise value for Iodine Software, which closed 1 October 2025. A combined $9–12B would make Datavant the single largest component of the $32B Thoreau package.

Datavant owns the health data supply chain and monetizes it per transaction:

• 70,000–80,000 connected hospitals and clinics
• 75% of the 100 largest US health systems
• 60M+ patient records exchanged annually; 30M+ records coded annually for payers
• 2026 Best in KLAS for Risk Adjustment Coding, Retrieval & Compliance, and for Outsourced Coding for payers
• Embedded in the Health Information Management department, through release of information

What it lacks is a high-margin product that reasons over the record it already moves. It is, structurally, a toll road.

SmarterDx is that engine. It reviews 100% of discharges pre-bill, identifies missing or incorrect diagnoses, and generates evidence-backed appeals. Its published economics: $2–3.5M of net new revenue per 10,000 discharges, 5:1 ROI, 30–50 basis points of net patient revenue, a KLAS satisfaction score of 98, and 100% client retention. Its catalogue spans SmarterPreBill, SmarterDenials, SmarterCharges and SmarterNotes, across clinical documentation integrity, charge capture and denials.

Four ways the two compose:

• Same building, different department. Datavant sells into HIM; SmarterDx sells into CDI and revenue integrity. Both monetize the same artifact — the chart. SmarterDx claims "60+ health system clients." Datavant reaches 75 of the 100 largest systems. Distribution is the trade.
• The payer side is the larger prize. Datavant already codes 30M+ records a year for payers and performs risk-adjustment retrieval. Reading a chart for diagnosis capture is computationally the same problem as HCC capture. Pointing SmarterDx's engine at Datavant's existing payer book opens a second market with no new customer acquisition.
• The AI layer is being assembled deliberately. Aetion for real-world evidence (July 2025), then Ontellus, then DigitalOwl for AI chart review in insurance and legal (October 2025), and now SmarterDx for provider-side chart reasoning. This is the fourth AI layer added to the retrieval network in roughly 15 months.
• Both sides of the table. Datavant retrieves records for payers, who use them to deny claims and adjust risk. It would now also sell providers the tooling to contest those denials. See section 7.

The immediate pool is mid-revenue-cycle leakage, and the buyer's pain is at a cyclical peak:

41% of providers now exceed a 10% denial rate, against 30% in 2022, and the average hospital absorbs roughly $5M a year in denial losses. SmarterDx prices as a share of recovered revenue, so its addressable market is a slice of that leakage rather than a seat count.

The competitive fight. Clinical Terminal's KLAS-segment index places the Clinical Documentation Integrity contest as follows. Tracked relationships are an ordinal traction signal drawn from a sparse mirror of public evidence — they are not audited customer counts.

The gap is the point. SmarterDx is loud but thin — the second-highest signal volume in the segment, on roughly a tenth of Iodine's tracked footprint. Iodine reaches more than 1,000 hospitals and now sits inside Waystar's payments platform. The target is therefore not greenfield; it is displacing Iodine and Waystar as the mid-cycle platform of record. SmarterDx cannot do that on its own distribution, which is precisely the argument for putting it behind Datavant's.

The strategic prize is the layer above both. Whoever owns the record pipe and the AI that reads it sets the terms of the payer–provider dispute over what a chart means. That is an infrastructure position rather than an application position, and it is the argument for a 15–20x multiple instead of an RCM services multiple. That multiple arbitrage is the deal thesis.

• Thoreau died in March. Proximate cause; the rest is supporting narrative.
• The comparable set repriced. Waystar paid $1.25B for Iodine in October 2025 and told investors it expanded their addressable market by more than 15%. Since then: R1 RCM agreed to acquire Humata Health on 18 August 2026; Med-Metrix closed Vitalware at $147M on 31 July 2026; Kyndryl agreed to acquire Healthcare IT Leaders on 10 August 2026. Becker's counted nine RCM M&A moves in the five months from 17 March.
• Denial economics deteriorated quickly, moving provider budgets toward revenue integrity.
• Provider AI spend is consolidating toward platforms. A $150M point solution inside a three-company services rollup is a weaker seller than the same product inside the company that already handles the customer's medical records.

Smarter Technologies launched in May 2025 with $800M+ in revenue, 200+ clients, 60+ health systems and Jeremy Delinsky, formerly CTO of athenahealth, as CEO. Removing its highest-multiple asset 15 months later implies one of three readings:

• The integration thesis did not hold. Access Healthcare is offshore business process outsourcing — labour arbitrage, services margins. SmarterDx is high-margin clinical AI. Different buyers, different margins, different multiples. Combining them suppressed SmarterDx's multiple rather than lifting the group's.
• SmarterDx is worth more as the growth engine of a levered Datavant than as one component of a services rollup.
• The assets are being re-cut for an exit vehicle, most plausibly a Datavant IPO.

All three carry the same implication: the May 2025 combination was portfolio construction rather than a product thesis, and it is being unwound for financing reasons. That is the most useful thing to hold in mind when the strategic rationale is eventually announced.

Stated plainly, because the numbers above should not be over-read:

• The transaction is unconfirmed. Single-source reporting, no company statement, no filing, contingent on a recap that has not closed.
• Customer relationships are largely logo-wall derived. Of SmarterDx's tracked relationships in our graph, only Novant Health carries a quotation that actually names the customer. The remainder are logo placements under "current clients include" — evidence of a claim, not of a contract.
• Facility-level duplication inflates raw counts. Three separate UCHealth rows and two Franciscan Health rows refer to the same systems at different facilities. They are not distinct customers.
• Datavant is under-represented in our segment view. It carries no product catalogue in our data and does not surface in the Release of Information segment at all, so segment rankings understate it by construction.
• Financial figures are as reported by third parties, not audited, and in the case of SmarterDx's revenue, disputed between sources.

A private equity liquidity problem wearing a strategic rationale. The rationale fits — pipe plus engine is the right architecture for mid-cycle AI, and Datavant's provider footprint is the one asset that could make SmarterDx genuinely competitive with Iodine and Waystar. But the transaction is happening because Thoreau collapsed in March, the vehicle is $2.5B of debt rather than a sale, and the growth asset being added consumes cash.

Two things to watch: whether the recapitalization actually closes, and whether Datavant's payer customers accept a supplier that now arms the other side of the denial fight.

Analysis prepared 21 August 2026. Company financials as reported by Axios Pro and other third-party outlets; relationship, product and segment data from Clinical Terminal.