AGX · Argan, Inc · Stock research

Current Thesis

Argan is the listed pure-play EPC contractor — through core subsidiary Gemma Power Systems — building the U.S. natural-gas generation that hyperscalers now contract for firm, 24/7 AI data-center load. The demand narrative is ACCELERATING: a ~$2.8B backlog anchored by 4.1+ GW of combined-cycle gas work, funded off $973.6M of cash and zero debt. The stock is a separate story. After the 2026-06-04 Q1 print gapped it to an all-time high of $805.75, AGX round-tripped the entire move — $630.32 close on 2026-07-10, $599.74 on 2026-07-14, $563.06 on 2026-07-25 — roughly 30% off the high and through both the ~$675 breakout shelf and the ~$600 round-number/Lake Street floor. Theme intact, price structure broken. The constructive read is a base-and-reclaim off the pre-run $500s, not a fresh chase into a falling tape ahead of the ~September Q2 backlog print.

Bullish and bearish views on Argan, Inc

The model's bull view on Argan, Inc (AGX), in brief: Q1 FY27 (reported 2026-06-04): record revenue $291.0M (+50.2% YoY); diluted EPS $3.24 vs $2.33 consensus; adjusted EBITDA +79% to $56.4M; gross margin 19.0%→21.0%. The bear view: Price has round-tripped: -30% from the $805.75 ATH, the June earnings gap fully filled, $563.06 close on 2026-07-25 sitting below both the ~$675 and ~$600 shelves — the momentum leg that was the setup has inverted into distribution. Both cases follow in full.

Bull Case

• Q1 FY27 (reported 2026-06-04): record revenue $291.0M (+50.2% YoY); diluted EPS $3.24 vs $2.33 consensus; adjusted EBITDA +79% to $56.4M; gross margin 19.0%→21.0%. Margin expanded on a record top line — operating leverage, not just volume.
• Backlog ~$2.8B at 2026-04-30, anchored by 4.1+ GW of gas-fired combined-cycle work (~79% gas mix), ~2.4x trailing revenue — multi-year visibility in the exact asset class signed for firm AI power.
• CPV Basin Ranch ~1,350 MW Texas EPC — full notice-to-proceed 2025-10-30, GE 7HA.03 turbines, completion 2028, plus a second ~860 MW ERCOT gas plant; the two added roughly $1B and ramp revenue recognition through FY27.
• Second growth leg: a ~$125M data-center project (thermal expansion + energy-storage tanks) and a new North Carolina fabrication facility completing later in 2026 for data-center pressure-vessel work.
• $973.6M cash and investments at 2026-04-30 (up from $895.0M at 2026-01-31), zero debt — growth and capital return are self-funded, with no equity-raise overhang into a hot tape.
• Capital return: dividend raised to $0.50/qtr (third straight hike); $200M buyback extended through 2030-01-31 (raised from $150M, 2026-04-08) — a stance inconsistent with a peaking order book.
• GE Vernova turbine slots sold through 2028 keep the downstream EPC funnel full; AGX awards lag GEV bookings by 2–4 quarters, so the OEM sell-out reads forward on Argan order flow.

Bear Case

• Price has round-tripped: -30% from the $805.75 ATH, the June earnings gap fully filled, $563.06 close on 2026-07-25 sitting below both the ~$675 and ~$600 shelves — the momentum leg that was the setup has inverted into distribution.
• Backlog dipped QoQ for the first time: $2.929B (2026-01-31) → ~$2.8B (2026-04-30), book-to-bill under 1 — revenue is now burning the book faster than awards refill it; the ~September Q2 print decides trend versus one-quarter air pocket.
• Still not cheap after the drawdown: ~$7.9B cap, trailing P/E ~48 — a second book-to-bill wobble at ~50x compresses the multiple even if the story holds; valuation is the downside vector here.
• Price now trades below the 5-analyst average target (~$679.80, as of ~2026-07-07) and below Lake Street's $600 Hold (raised from $375, 2026-06-05) — a month ago it was above every target; the sell-side air cover is gone.
• Top-customer concentration >60% per the 10-K risk factors — a single gas-plant delay produces outsized single-day gaps; sizing discipline is mandatory regardless of conviction.
• The ~13% renewable backlog sleeve carries IRA/policy exposure into a shifting subsidy environment; only the ~79% gas sleeve is the clean thesis.

Setup & Price Structure

The name is mid-correction inside an accelerating theme, and the divergence is the whole problem. From the $805.75 ATH, AGX has given back the entire post-earnings advance and lost two reference shelves: the ~$675 June breakout base and the ~$600 round number that doubled as Lake Street's target. At $563.06 (2026-07-25) the next real reference is the pre-run consolidation in the $500s — the zone AGX built before the Q1 pop. Below that, the 52-week structure thins toward the $400s; the $196.90 52-week low is a pre-supercycle artifact, not usable support. Trend, the breadth of the give-back, and the round-number break line up the same way: mean reversion in progress, six weeks ahead of a binary print. The setup that pays is a higher low in the $500s followed by a reclaim of $600 on expanding volume. Until then, the strength that defined the original entry has left the tape. Stand aside until it bases.

Catalyst Calendar (next 30 days)

• ~2026-08-27 (est.): quarterly dividend declaration — watch for a fourth consecutive hike off $0.50/qtr; a pause would be an early signal on management's book confidence.
• ~2026-09-03 (est.): Q2 FY27 earnings (quarter ends 2026-07-31) — outside the 30-day window but the next binary: backlog trajectory (rebuild above the $2.9B January peak vs a second straight decline toward ~$2.5B) and gas-mix commentary. Avoid fresh entries within three trading days of the confirmed date.

Elapsed catalysts

• ~2026-07-29 to ~2026-08-01 (est.): hyperscaler Q2 2026 capex prints (MSFT, META, AMZN). These set the FY27 data-center capex tone that drives the entire gas-EPC funnel; a >10% FY27 capex cut from a top-4 buyer is the theme's largest near-term risk, an upside guide its largest tailwind. Not an AGX report, but the dominant near-term driver of the tape. (passed 8d ago)
• ~2026-07-23 (just passed): GE Vernova Q2 2026 print (est.) — turbine-order and lead-time commentary is the forward read on Argan's award pipeline (2–4 quarter lag). Track GEV's post-print structure and 20-EMA as the leading indicator. (passed 17d ago)

What Would Change Our Mind

A lower price alone does not flip the read constructive. What would: a higher low holding the $500s, then a weekly reclaim of $600 on expanding volume, ideally with the ~September Q2 print rebuilding backlog back above the $2.9B January peak (book-to-bill back over 1). A GEV print showing accelerating turbine bookings and extending lead times would corroborate the forward funnel. The thesis breaks instead on a weekly close below $500 — that loses the pre-run consolidation base and argues the supercycle re-rate is unwinding toward the $400s rather than basing. A second consecutive QoQ backlog decline toward ~$2.5B (book-to-bill under 1 for two straight quarters), or a top-4 hyperscaler cutting FY27 capex guide by >10%, would confirm the demand narrative itself has cracked beyond the price. Mainstream "data-center power crunch" cover-story saturation arriving while spot already sits below every analyst target would mark late-cycle retail catching a move that has already turned.

Correlation Notes

AGX trades as a high-beta derivative of GE Vernova (GEV): Argan builds the plants around GEV turbines, and its backlog follows GEV bookings 2–4 quarters downstream, so GEV's 20-EMA and order commentary lead AGX. The name is tethered to the broader AI-power complex (GEV, VST, CEG, TLN, NRG, PWR) and, one layer up, to hyperscaler capex intentions (MSFT, META, AMZN, GOOGL) — the demand that fills the EPC funnel. The ~13% renewable backlog sleeve adds IRA/clean-energy policy beta the gas sleeve does not carry. As a long-duration growth infrastructure name at ~50x, AGX de-rates on rising-real-yield and tightening regimes independent of company news. Beta to the AI-compute leaders (NVDA) is real but second-order: power is a downstream tell on AI-capex durability, so AGX tends to confirm, then lag, moves in the compute layer.