Innventure Reroutes Accelsius: Milestones Over Revenue as Two-Phase Cooling Waits on Big Tech

The company-builder suspends Accelsius revenue guidance, abandons its 2028 cash-flow-positive goal, and bets two-phase direct-to-chip cooling on hyperscaler patience — after shares shed half their value in eight days.

INV · Earnings Call · 2026-08-13

Innventure's second-quarter call was less an earnings report than a strategic confession. The company-builder that had guided Accelsius to an ~$100 million annualized revenue run rate by year-end and to consolidated positive cash flow in 2028 scrapped both. Shares responded brutally: the stock has fallen about 55% in the eight trading days since the August 13 report, extending a 90-day drawdown that has wiped out nearly 80% from the May 14 peak.

Guiding on Milestones, Not Revenue

The most consequential change is philosophical. CFO Dave Yablunosky announced a suspension of forward guidance:

Due to these structural constraints, which limit <keyword id="eedd67d5ea">early adopter</keyword> deployments, we are suspending our revenue targets.
David Yablunosky, Chief Financial Officer · 2026-08-13

Incoming CEO Bill Grieco, an engineer by training, framed it as a governance shift: “predicting revenue quarter-by-quarter is difficult, if not impossible... revenue targets are not the most useful yardstick at this stage.” — William Grieco, Incoming Chief Executive Officer · 2026-08-13 This pivot replaces revenue targets with four key milestones — chip-maker engagement, server OEM/ODM co-development, a hyperscaler statement of work, and benchmark data with leading thermal labs. Most companies narrow guidance; Innventure abandoned it entirely.

Two-Phase Physics vs. Market Structure

Underneath the guidance change is a market-structure argument. Accelsius makes two-phase direct-to-chip cooling, and new CEO John Hewitt (ex-Vertiv Americas) argues the physics is inevitable: “Adopting this technology requires GPU allocations, access to power and the scale to influence server designs, and those are precisely the things smaller companies can't get.” — John Hewitt, Chief Executive Officer (Accelsius) · 2026-08-13 The net effect:

in this market, there are very few <keyword id="01fc6e2473">adopter</keyword>s.
John Hewitt, Chief Executive Officer (Accelsius) · 2026-08-13

The same GPU and power constraints that strand smaller early adopters concentrate purchases among hyperscalers — and those evaluation cycles run two to three years. So Hewitt is courting the behemoths directly, with chip makers, server OEMs/ODMs, and hyperscalers as the four adoption pillars. The company's technological conviction rests on a July third-party benchmark that retrofitted a standard Dell PowerEdge server (8-way NVIDIA B200) with its NeuCool cold plates. Results: GPUs ran 9–14°C cooler at the system level, at ~1/3 the chip coolant flow, and held 9°C of headroom at 50°C facility water — where the single-phase baseline throttled at 84°C. That warm water readiness is the energy story: fixed power envelopes mean every watt saved on cooling is a watt redirected to compute, and Hewitt cites a Jacobs reference design showing ~5% more GPUs per fixed power. The team deliberately positions this against the dominant incumbent — single-phase cooling tops out around 45°C facility water.

We didn't hand them a marketing claim. We handed them a data set.
John Hewitt, Chief Executive Officer (Accelsius) · 2026-08-13

The DarkNX Setback and the Capital Question

The concrete casualty of these constraints is the previously touted DarkNX order. Yablunosky explained: “The purchase order referenced a development site that DarkNX had previously identified. DarkNX recently informed Accelsius that this site is no longer available” — David Yablunosky, Chief Financial Officer · 2026-08-13 — and in Q&A, Hewitt confirmed the site was lost, in his words, “to the power envelope” — John Hewitt, Chief Executive Officer (Accelsius) · 2026-08-13. The order was debooked and removed from the 2026 forecast, though it may transfer to an alternate site once one is secured. This sharp reversal is worth contrasting with management's tone three months prior. In May, Bill Haskell was unequivocal about the $100M run-rate target: “We still feel very optimistic that we'll hit that run rate before year-end... no reason to believe that we're not going to hit that target.” — Gregory Haskell, Chief Executive Officer · 2026-05-14 He also promised “by the time we announce the next quarter, which will be first half of August, in all likelihood, we can give an update on bookings” — Gregory Haskell, Chief Executive Officer · 2026-05-14. None of that — the run rate or the bookings update — has materialized. Financially, the reset matters. Free cash flow (less SBC) widened to a -$40M loss in Q2, and year-to-date operating cash use reached $59.5M. Yablunosky insists Innventure still holds $41.5M cash and retains the standby equity purchase agreement, but he conceded: “there could be a need for cash to do additional cash raises in the second half of the year.” — David Yablunosky, Chief Financial Officer · 2026-08-13 AeroFlexx and Refinity are expected to raise at the operating-company level, protecting Innventure's pro-rata exposure to Accelsius — but the parent still faces dilution. Innventure is making a portfolio-wide bet that two-phase cooling's adoption curve, currently bottlenecked by GPU and power scarcity, will reward patience. The leadership transition — Haskell to Grieco, with Hewitt taking Accelsius — is designed to execute that bet. But the market is voting with its feet: even after the 90-day decline, the price-to-sales multiple sits near 95x on negligible revenue, so the entire thesis rests on hyperscalers converting benchmark data into statements of work before the cash runs out.