Plug Power (PLUG) — P&L Turnaround & Path to EBITDAS

Reading the chart. The margin line is the story: five of the last six quarters improved sequentially, and the two that broke the pattern (Q3'25, and the Q4'24 trough) were driven by inventory and equipment-cost charges, not demand. Revenue mix is shifting under the surface — equipment was 57% of revenue in Q1'24 and is 46% in Q2'26, while recurring streams (services, PPA, fuel) grew from 41% to 54%. Guidance math implies a steep second half: 15–16% FY growth on $709.9M of FY25 revenue means ~$475–480M in H2 vs $341.8M in H1 — management attributes this to the historically second-half-weighted cadence and the commercial backlog (Orica 50 MW FID, Carlton Barrow 30 MW, Hy2gen 275 MW FEED).

Three different stories. Equipment swung from (98)% (Q1'24) to +1.9% — the mix shifted away from loss-leading electrolyzer/cryogenic deliveries and Q4'24/Q3'25 absorbed the inventory writedowns that made old quarters look worse. Services (shown ex-provision) reached +27% in Q2'26, the margin management cites — reliability gains raised units-per-technician, and the expanding GenDrive installed base (76,000+ units) turns each refresh cycle into aftermarket revenue. Fuel remains the structural loss-maker at (48)%, though halved from (91)% a year ago on plant utilization (Georgia/Tennessee/Louisiana network, ~40 TPD). PPA at (30)% improves slowly as older, mispriced contracts roll off.

Methodology. Services margin excludes the "provision for loss contracts" line — a reserve true-up for future-period contracts that has recently been a large benefit ($15.7M released in Q2'26 alone). Including it would show services at +80%, which flatters current-period economics; excluding it may still understate historical losses. The provision path is visible in the table below.

The halving, audited. Reported Q2'26 opex of $62.4M is ~50% below Q2'25's $123.5M — but $19.4M of impairment sits in that figure, and more importantly the quarter benefited from a $39.7M "recovery of previously impaired assets" (disclosed in the adjusted-EPS reconciliation; the GAAP line it sits in is not disclosed — the $40.9M sequential drop in SG&A to $29.3M strongly suggests SG&A, which would put underlying SG&A near $69M, our inference). On that basis the underlying opex decline is roughly 20% YoY, not 50% — still real (Project Quantum Leap completed Q4'25: workforce, footprint, vendor renegotiation), but the headline overstates it. Kitchen-sink impairments remain a Q4 tradition: $940.9M in Q4'24, $666.3M in Q4'25, mostly against the hydrogen build-out.

What Q4'26 EBITDAS-positive requires. Core operating loss (ex impairment, restructuring, contingent consideration) narrowed from $262M in Q1'24 to $44M in Q2'26. EBITDAS adds back D&A and stock comp — about $20M per quarter at the H1'26 run rate — so the implied EBITDAS gap in Q2 was roughly $24M... before crediting back the $39.7M one-time recovery, which would swing the underlying gap to ~$64M. Closing that by Q4 leans on three levers moving together: the guided H2 revenue ramp (~$240M+/quarter vs $178M in Q2), continued gross-margin gains (Q4'25 already printed +2.4% on $225M of revenue — seasonally strong quarters do clear zero), and service/fuel scale effects. The target is plausible on a seasonally-strong Q4 with the revenue guide intact; it is not yet a run-rate statement.

Methodology note. Q1–Q3 figures are standalone-quarter columns from the respective 10-Qs; Q4 columns are derived as FY (10-K) minus nine-month YTD, so Q4 absorbs any full-year reclassifications — the Q4'25 services-revenue figure ($41.5M vs $19.7M in Q3'25) partly reflects a 10-K reclassification rather than pure sequential growth, and Q4 carries the annual kitchen-sink impairments. Net loss (Plug) is net loss attributable to Plug Power Inc.; it includes large non-cash fair-value swings on the 2033 convertible notes and $7.75 warrants ($103.5M combined loss in Q2'26) that sit below the operating line.