AEIS · Advanced Energy Industries, Inc.

Current Thesis

Advanced Energy is mid-re-rating from a cyclical semiconductor/industrial power vendor into a picks-and-shovels supplier of AI-datacenter power conversion, and after a sharp early-June shakeout the tradable structure has repaired. The fundamental leg is accelerating: Data Center Computing revenue doubled to a record $194.2M in Q1 (reported 2026-05-04) on top of +107% in 2025, and the 2026-06-03 launch of the ADH 800V DC-DC converter series specs the product directly into the 800V HVDC rack architecture NVIDIA is pushing for Rubin/Kyber-class megawatt racks. Price flushed to the 200-day MA (~$286) into 2026-06-05/06, then rallied roughly 22% in five sessions to reclaim the 50-day MA (~$346), closing near $354 on 2026-06-12 (+4.1%) after Cantor Fitzgerald initiated Overweight, $400 on 2026-06-11 (~8.4% intraday). The narrative an investor is buying here: the AI buildout is power-constrained, and AEIS sells the conversion hardware inside every 800V rack. The catch is that the bounce is extended into no scheduled catalyst before the ~early-August Q2 print, while a fresh $1.15B 0% convert is delta-hedged short into rallies.

Bull Case

• DC segment compounding, not a one-print spike: Data Center Computing revenue +107% in 2025, then doubled again to a record $194.2M in Q1 2026 (call 2026-05-04); FY26 DC-segment growth guided to mid-30s%.
• Q1 beat-and-raise (2026-05-04): total revenue $511.0M (+26.3% YoY); EPS $2.09 (+69.9% YoY), ahead of consensus. Q2 guide $520–560M (mid ~$540M vs ~$527M est) and adj EPS $1.93–$2.43; FY26 total-revenue growth lifted to low-to-mid-20s%.
• Fresh hardware proof of the 800V thesis (2026-06-03): ADH series converts 800V→50V at 98.2% peak efficiency, 8kW peak / 6kW continuous in a half-brick at >2,700 W/in³, liquid-cooled and parallelable; with the NDQ 50V→12V stage and a Hot Swap Control module it forms a complete 800V rack solution aimed at NVIDIA Rubin-era designs.
• New sell-side coverage adds a fresh buyer (2026-06-11): Cantor Fitzgerald initiated Overweight, $400, citing semi-equipment spend, datacenter power, and recovering industrial end markets sparking an ~8.4% session.
• Structure repaired: the 2026-06-05/06 test of the 200-day MA (~$286) held, and the ~22% rebound reclaimed the 50-day MA (~$346), reversing the broken-tape condition that defined the early-June read.
• Cap structure termed out (2026-06-12): notice to redeem all remaining $136.7M of 2.50% converts due 2028 on 2026-09-23, completing the refinancing into the longer-dated 0% 2031 notes lower coupon, higher conversion reference, maturity pushed to 2031.

Bear Case

• The chase is extended: ~+22% in five sessions into the 50-day reclaim leaves price stretched short-term with no scheduled catalyst for ~7–8 weeks, and the consensus average target sits only ~$358 (Moderate Buy), so spot near $354 has already closed most of the gap to the average estimate.
• Sell-side is largely published: the 2026-05-05/07 cluster (BofA $430, Citi $410, Susquehanna $430, KeyBanc $415, Needham $400, Baird $380, plus bears TD Cowen $350 / Wells $345) plus Cantor's $400 means the narrative is mainstream; the upgrade-driven re-rating fuel is mostly spent below the ~$430 top end.
• Convert-arb overhang is mechanical and live: the $1.15B (incl. $150M greenshoe) 0% notes due 2031, priced 2026-05-13/14 at a ~50% premium to $339.19 (~$508.79 conversion), are delta-hedged by arb desks shorting the equity into strength structural supply on every rally.
• cash settles the principal but the excess-over-par is delivered in shares.
• Single-engine risk: management noted on the 2026-05-04 call that DC strength is masking still-soft semiconductor-equipment demand; a DC air-pocket before semi recovers would remove the only growth driver.
• Valuation is full: third-party value screens flag the stock as overvalued after a +~70% YTD run, against a 52-week range of roughly $122–$397.

Setup & Price Structure

The early-June break looked terminal a -8.1% session on 2026-06-05 undercut the $298.85 June low and parked price on the 200-day MA (~$286) but that line held and the snap-back has been violent: roughly +22% into 2026-06-12, reclaiming the 50-day MA (~$346) and closing near $354. The week ending 2026-06-12 therefore printed a weekly close back above the 50-day, the condition that flips the structure from broken to repairing. Overhead resistance is the April highs near $387–$397 (52-week high ~$397.44); support steps down to the reclaimed 50-day (~$346), then the 200-day (~$286). The cleanest continuation entry is a higher-low retest that holds the 50-day rather than a chase right after a near-vertical 22% move; a volume breakout through $387–397 would be the other add trigger. RSI is elevated after the run, which in an accelerating, cluster-confirmed power name is confirmation rather than a reason to fade but it does cap the reward on a fresh entry at $354 versus the ~$286 test eight sessions earlier.

Catalyst Calendar (next 30 days)

• No scheduled hard catalyst before mid-July. The next binary is the Q2 2026 print, expected ~early August (not yet scheduled) outside this window.
• 2026-06-12 (done): redemption notice for the remaining $136.7M 2.50% 2028 converts; conversion window runs to 2026-09-22, redemption 2026-09-23 a cap-structure cleanup with little near-term tape impact.
• Ongoing, unscheduled: ADH/800V design-win or NVIDIA reference-design announcements. A named hyperscaler/OEM 800V socket win would be the next narrative accelerant and the trigger to re-rate conviction up.
• Ongoing: convert-arb delta-hedging on the 0% 2031 notes mechanical supply into rallies, not a dated event.

What Would Change Our Mind

• Bull-confirming: a weekly hold above the 50-day (~$346) that builds a higher-low base, or a volume breakout through the $387–397 highs, would convert the bounce into a fresh trending leg and justify higher conviction.
• Bear-confirming (structure): a weekly close back below the reclaimed 50-day MA (~$346) negates the rebound and reopens the broken $286–346 range; a weekly close below the 200-day (~$286) ends the 2026 uptrend and marks the name a value trap not a level to average into.
• Bear-confirming (fundamentals): Q2 (~early-Aug) Data Center Computing revenue printing a sequential decline versus Q1's $194.2M record, or any cut to the FY26 mid-30s% DC-growth guide, breaks the core thesis regardless of price.
• Catalyst awareness: avoid fresh entries inside three trading days of the Q2 date once it is scheduled the print is binary and the thesis is not earnings-timed.

Correlation Notes

AEIS trades as a high-beta proxy for two baskets: AI-datacenter power/cooling infrastructure (Vertiv/VRT, Eaton/ETN, GE Vernova/GEV) and semiconductor-equipment capex (Applied Materials, Lam, KLA / WFE spend), with the DC narrative currently driving the tape and the legacy SEMI segment the slower-moving anchor. The dominant single-name driver is NVIDIA's datacenter roadmap: the 800V HVDC architecture for Rubin/Kyber megawatt racks is the demand thesis, so AEIS reacts to NVIDIA power-architecture and Rubin-timeline headlines. Convert-arb on the 0% 2031 notes adds an idiosyncratic mechanical short into rallies that can decouple the equity from the theme on up days. As an industrial with semi exposure it is also rate- and PMI-sensitive a hawkish macro turn or a renewed semi-equipment air-pocket pressures the multiple independent of the AI-power story.