AEIS · Advanced Energy Industries, Inc. · Stock research

Current Thesis

The 2026-08-03 Q2 print resolved the tension the previous note flagged — analyst targets climbing while price carved lower lows — in favour of the numbers. Revenue came in at $574.1M against $542.8M consensus, adjusted EPS $2.74 against $2.20, and management raised the FY26 revenue-growth guide to low-to-mid-30% from low-to-mid-20% while lifting the Data Center Computing growth guide to 50%+ from mid-30s. Q3 is guided to roughly $640M ±$20M ($620–660M) versus $573.7M consensus, with non-GAAP EPS $3.00 ±$0.25. The tape has not paid for that: the 2026-08-14 close of $330.26 sits 15.1% below the 52-week high of $388.93, and the three-month return is +2.1%. The leg being bought is a two-engine capital-equipment vendor — a record $278M semiconductor quarter plus $192M of AI datacenter power conversion — where the estimate line is rising faster than the price line. What that leg does not yet contain is 800V revenue: on the Q2 call management placed high-volume production of 800V data-center product in 2028.

Bullish and bearish views on Advanced Energy Industries, Inc.

The model's bull view on Advanced Energy Industries, Inc. (AEIS), in brief: Q2 beat on both lines (2026-08-03): revenue $574.1M (+30% YoY) vs $542.8M consensus; adjusted EPS $2.74 (+83% YoY) vs $2.20 consensus; operating income a record $125M at a 21.9% operating margin, +730bps YoY. The bear view: The AI segment went sideways sequentially: Data Center Computing was $192M in Q2, -1% QoQ from the $194.2M record set in Q1 (reported 2026-05-04). Both cases follow in full.

Bull Case

• Q2 beat on both lines (2026-08-03): revenue $574.1M (+30% YoY) vs $542.8M consensus; adjusted EPS $2.74 (+83% YoY) vs $2.20 consensus; operating income a record $125M at a 21.9% operating margin, +730bps YoY.
• The raise was larger than the beat (2026-08-03 call): FY26 revenue growth guided to low-to-mid-30% from low-to-mid-20%, and the Data Center Computing FY26 growth guide moved to 50%+ from mid-30s on accelerated hyperscale investment. Q3 revenue guidance of $620–660M is roughly $50–85M above the $573.7M the Street carried.
• Semiconductor is carrying the quarter, not just AI: the segment printed a record $278M, +33% YoY and +27% QoQ, with management pointing to nearly 50% YoY H2 semiconductor growth — the WFE-cyclical half of the business is expanding at the same time as the datacenter half.
• Margin structure improved with the mix: gross margin 41.9% in Q2 (40.7% excluding tariff refunds), a 380bps YoY improvement, with Q3 guided at 41–41.5% on opex of $120–124M.
• Balance sheet funds the ramp: $1.4B cash and $132M net cash at Q2, FY26 capex guided $180–195M, Thailand factory first production revenue expected in Q4 2026, and management citing more than $2.6B of current revenue-generating capacity against ~$5B once the global network is fully operational.
• Post-print target round was broad (2026-08-04/08-05): Citi to $460, Morgan Stanley to $439, Needham reiterating $400, Seaport Global upgrading to Buy at $410. Even TD Cowen, still at Hold, raised to $370 — above the 2026-08-14 close.

Bear Case

• The AI segment went sideways sequentially: Data Center Computing was $192M in Q2, -1% QoQ from the $194.2M record set in Q1 (reported 2026-05-04). The 50%+ FY26 segment guide is therefore a second-half-weighted claim that has not yet been demonstrated in a printed quarter.
• The 800V story is a 2028 revenue event: the ADH 800V→50V converter launched 2026-06-03, but management guided high-volume production of 800V data-center products to begin in 2028 on the Q2 call. An investor buying the HVDC rack narrative today is funding three intervening quarters of hyperscaler power shelves and semiconductor tools.
• Coverage is already fully deployed: after the May cluster (BofA $430, Susquehanna $535 on 06-30, Wells Fargo $465 on 07-01, Citi $435 and Morgan Stanley's Overweight initiation at $421 on 07-22), the 08-04/08-05 round was mostly incumbents raising numbers. Of 12 analysts tracked by stockanalysis.com in mid-August, the average 12-month target sits near $428.73 — roughly 30% above the 2026-08-14 close, with the marginal upgrader already spent.
• Concentration risk on both engines: management describes focus on "select" hyperscale customers with a "second wave" of datacenter customers only reaching parity with the largest hyperscaler by 2028–2029; the semiconductor segment, at $278M of the $574.1M quarter, is levered to a WFE cycle that has historically turned faster than the datacenter one.
• Convert overhang persists: the $1.15B 0% notes due 2031 priced 2026-05-13/14 with a conversion reference near $508.79 leave arbitrage delta-hedging as a structural supply source into rallies.
• The July damage is recent: the stock fell 12.65% on 2026-07-02 to close $311.27 and reached $287.73 on 2026-07-07. Those levels are two months old, not two years old.

Setup & Price Structure

The narrative is maturing. The narrative is fully known — ten-plus banks published on it, and Morgan Stanley's initiation on 2026-07-22 was the last major desk to arrive — but it is still working on fundamentals, evidenced by the 2026-08-03 beat and the FY26 guide raise, and flow has moderated rather than stopped. It is not accelerating: a +2.1% three-month return and a close 15.1% below the 52-week high is not expanding participation. It is not saturated either, because the earnings revisions running through 08-04/08-05 were genuinely upward and price has not capitalised them.

Structure as of the 2026-08-14 close of $330.26: price sits inside the July–August range, above the 2026-07-07 low of $287.73 and well under the 52-week high of $388.93. RSI(14) at 60.1 is mid-range — the post-print move did not produce an overbought tape, which is consistent with a stock digesting a guide raise rather than chasing it. The zone that matters on the downside is the pre-print July shelf around $311, the 2026-07-02 close; giving that back means the Q2 beat bought nothing durable.

Crowding and positioning observables, stated as observables: (1) the analyst target distribution is clustered $370–$460 after 08-04/08-05, with the $535 Susquehanna outlier from 06-30 still standing — an average near $428.73 across 12 analysts; (2) retail-facing coverage has shifted to backward-looking performance content, with Benzinga's "$100 invested 20 years ago" piece on 2026-08-07 citing a 17.29% annualised return; a small, pre-scheduled disposal, not a discretionary exit; (4) no earnings event sits inside the next 30 days, so the near-term calendar is conference headlines rather than a print.

Catalyst Calendar (next 30 days)

• 2026-09-10, 8:50–9:25 AM ET — Citi Global TMT Conference presentation (webcast on ir.advancedenergy.com). Citi carries a $460 target set 2026-08-04; the question that matters is any update on hyperscaler design wins, the second-wave customer set, or the 800V production timeline.
• 2026-09-23 — scheduled redemption of the remaining $136.7M of 2.50% convertible notes due 2028, completing the refinance into the $1.15B 0% notes due 2031 (announced 2026-06-12). A mechanical, dated event: it removes the 2028 coupon and shifts the remaining conversion overhang to the ~$508.79 reference.
• ~2026-11-02 (est.) — Q3 print, first test of the $620–660M revenue guide and the 50%+ Data Center Computing FY26 growth claim. Not inside 30 days; it is the next binary.

Elapsed catalysts

• 2026-08-18 — Seaport Research Partners Annual Summer Investor Conference appearance per the company's investor-events listing; Seaport upgraded to Buy at $410 on 2026-08-05, so the fireside is the first public airing of that thesis with management. (passed 8d ago)
• Q4 2026 (no fixed date) — first production revenue from the Thailand factory, per the 2026-08-03 call. (passed 23d ago)

What Would Change Our Mind

The structure that has to hold is the post-Q2 shelf. The 2026-08-03 beat and guide raise are only worth something if price keeps them: losing the pre-print July range means the market has decided the raise is either not credible or already discounted. Concretely, a weekly close below $311 gives back the entire post-print gap and returns the chart to the July regime that produced the 12.65% single-day loss on 07-02; a weekly close below $287.73 would take out the July low and break the multi-quarter uptrend outright.

On fundamentals, three things would flip the read. First, Data Center Computing printing below $192M again at the Q3 report — a second consecutive sequential decline would make the 50%+ FY26 segment guide arithmetically hard and would sit badly against a story sold on AI power. Second, Q3 revenue landing under the $620M low end of guidance, which would be the first guidance failure of this cycle. Third, any slip in the 2028 high-volume 800V timeline at the 09-10 Citi appearance or the Q3 call — the HVDC leg is the part of the narrative with no revenue attached yet, so it is the part that can be repriced on words alone. Conversely, an unscheduled 800V design-win or a named hyperscaler reference announcement is the datapoint that would move this back toward accelerating.

Correlation Notes

• Roughly 48% of Q2 revenue ($278M of $574.1M) came from the Semiconductor segment, so the name tracks the WFE complex — AMAT, LRCX, KLA, MKSI — as much as it tracks AI power. A WFE order digestion signal from any large tool OEM hits the larger of the two engines.
• The remaining datacenter beta (~33% of revenue, $192M) moves with electrical-infrastructure and rack-power names (VRT, MOD, ETN) and with the NVIDIA rack roadmap, since the 800V architecture the ADH series targets is specified around Rubin/Kyber-class designs.
• The $1.15B 0% convertible due 2031 introduces a non-fundamental flow: hedged holders sell delta into strength, which can cap upside moves that would otherwise extend on news.
• Guidance is dollar-denominated with a tariff-refund component in the Q2 gross margin (41.9% reported vs 40.7% excluding refunds), so trade-policy headlines feed the margin line directly and not only through demand.