Aehr Test Systems has production orders supporting an AI burn-in expansion, but its latest reported GAAP net profit does not establish operating profitability. Our initial $37 Base value requires a substantial operating recovery and remains below the $63.88 regular-session reference price. The central uncertainty is whether concentrated customer ramps can produce durable margins and cash per share after working capital and dilution.
AHNIM STUDIO · Company Research · Nasdaq: AEHR. Valuation date: October 22, 2026. Target date: approximately October 22, 2027, using FY2028 operating scenarios. Financial amounts are consolidated USD millions unless stated; targets are USD per share. This is the initial baseline, model AHNIM-AEHR-20261022-R1.
The market reference is the October 21, 2026 regular-session close at 4:00 p.m. EDT, equivalent to October 22 at 5:00 a.m. KST: $63.88. The provider reports approximately $2.08 billion of market capitalization, below the original $10 billion small-cap screening ceiling. That is an approximate vendor figure using a dated share count; it is not a newly verified October share register. The stock remains quoted on Nasdaq, with no completed acquisition or delisting identified in the reviewed official material.
1. How the business earns revenue
Burn-in subjects semiconductors to sustained electrical and thermal stress to screen for early failures. Aehr sells the equipment and device interfaces that perform that work. Its opportunity is tied to manufacturing reliability and the cost of a failed device, rather than to a royalty on every AI computation.
The FOX-XP wafer-level offering screens devices before they are cut from the wafer and incorporated into an expensive package. Its August order includes systems, proprietary WaferPak contactors and automated alignment. Sonoma package-level systems address assembled high-power processors with burn-in modules, sockets and software. The two stages can serve different customer test strategies. Selling both does not mean every chip needs both steps or that Aehr captures both automatically.
| Revenue engine | Economic attraction | What can disappoint |
|---|---|---|
| Systems and automation | Customers buy test capacity for production programs | Large orders are lumpy and installation schedules can move |
| Contactors, modules and sockets | Device-specific interfaces can follow installed capacity or a new design | Replacement and redesign purchases are not a fixed subscription |
| Service, upgrades and engineering | Support can deepen an installed relationship | Engineering effort and field support consume resources |
The useful commercial loop is qualification → production selection → systems and interfaces → accepted delivery → repeat orders and collections. Once a test recipe and interface work in a customer’s process, switching can impose requalification effort. That creates potential friction, but it does not prove exclusivity, a guaranteed consumables annuity or permanent pricing power.
2. What the orders prove—and what the backlog cannot tell us
| Evidence date | Verified company disclosure | Boundary |
|---|---|---|
| July 14 results release | May 29 backlog $80.6m; effective backlog $100.6m includes subsequent bookings | Two different cutoffs, not a fresh October net backlog |
| August 12 issuer announcement | $22m follow-on FOX wafer-level AI order; six-month expected shipment schedule | A production order, with delivery and recognition still to follow |
| October 7 issuer announcement | Approximately $6m Sonoma follow-on orders for the same hyperscaler’s second processor | A second chip program, not a second customer; six-month deliveries and expected calendar-2027 ramp |
The July results, August FOX order and October Sonoma order strengthen the production-demand evidence. They cannot be summed into a current backlog: intervening shipments, recognized revenue, cancellations, rescheduling and other orders would all be needed for that roll-forward. Nor should the October order be assigned wholesale to the next quarter.
Aehr has not named the lead customers in these announcements. We do not identify an unnamed hyperscaler as Google, Amazon, Microsoft, Meta or NVIDIA. The manufacturing partner’s location and an application description are insufficient evidence. Diversification requires additional independent customer revenue, not merely more devices from one account.
The June silicon-photonics production order adds another demand path: reliability testing for optical devices used in data-center interconnects. More optical bandwidth can increase the importance of a reliable device population, but the conversion into Aehr sales depends on the customer’s test flow, selected supplier and capacity purchases. A wider industry market is not Aehr’s contracted market share.
3. Competition and the customer’s cost-of-test decision
AEM’s AMPS-BI announcement and Micro Control’s AI/HPC burn-in offering establish competing high-power solutions. Aehr’s narrower proposition is the combination of wafer-level interfaces and automation with its packaged-device portfolio. We have not verified a comparable market-share estimate or an independent head-to-head cost benchmark.
A buyer compares throughput, thermal uniformity, electrical stress coverage, uptime, handling labor, interface cost and the value of failures caught early. Detecting a defective die before adding costly memory and packaging can avoid downstream waste. The benefit still has to exceed the additional screening cost and cycle time. Better chip yields or a different reliability strategy can change that trade-off.
The same logic explains why an AI spending boom need not produce proportional equipment revenue. Once enough test capacity is installed, higher utilization can support more chip output before another system is needed. Conversely, a more demanding test recipe can require longer residence time and more capacity at unchanged unit volumes. Test intensity and utilization matter alongside processor shipments.
4. Financials: the profit headline and the operating result diverge
| FY2026 reported measure | USD million, except margin |
|---|---|
| Revenue | 50.001 |
| GAAP gross margin | 35.3% |
| GAAP operating loss | (14.149) |
| GAAP net loss | (7.126) |
| Operating cash flow | (3.310) |
| Property/equipment purchases | 2.066 |
| Cash flow less those purchases — calculated | (5.376) |
| Cash and equivalents, May 29 | 116.358 |
| Restricted cash, May 29 | 0.150 |
| Net public-offering cash proceeds | 97.395 |
| Contactors revenue: FY2025 → FY2026 | 30.848 → 14.887 |
Three anonymous customers supplied 26.3%, 14.2% and 10.9% of FY2026 revenue, a calculated 51.4% combined. The May balance sheet reported no financial borrowings and $9.882m of operating lease liabilities. [1]
Our interpretation is that the cash buffer reduces immediate financing pressure, while its origin matters: equity issuance funded the increase. The contactor decline also demonstrates that an installed base does not guarantee recurring sales. A full free-cash-flow assessment must additionally consider acquisitions and other investment uses.
In Q4, GAAP operating loss was $1.206m and pretax loss was $0.455m; a $1.846m tax benefit produced $1.391m net income. Full-year adjusted net income of $0.910m also coexisted with an adjusted operating loss of $6.152m. The reconciliation therefore does not support calling the company operationally profitable for the year. Tax benefits and interest are not recurring equipment margins.
The commercial recovery must show up first in gross profit after assembly, warranty and delivery costs, then in operating profit after engineering, selling costs and stock compensation. It must also survive the inventory and receivables needed to fulfill large orders. Cash received as a deposit helps funding but does not by itself complete revenue recognition or establish final program profitability.
5. The reporting calendar prevents a false “latest quarter”
The fiscal-year change creates a 28-day transition period from May 30 through June 26, 2026. The April 7 Form 8-K says transition reporting will be included with the next first-quarter filing. It should not be silently added to the May year or the ordinary first quarter.
The October 8 official calendar defines Q1 FY2027 as June 27–September 25 and FY2027 as ending June 25, 2027. Results are scheduled for November 2 after the U.S. close, with a 5:00 p.m. Eastern call—November 3 at 7:00 a.m. KST. As of this review, that report remains ahead; there is no verified September balance sheet or Q1 result in this analysis.
Management’s July outlook for FY2027 is revenue of $130m–$150m and non-GAAP net income of 18%–22% of sales. [2] Our own FY2028 GAAP operating-margin assumptions below measure a different period and accounting line. The company outlook cannot be inserted directly as an achieved operating margin.
6. Dilution: use economic claims rather than loss-period diluted EPS
The definitive proxy reports 32,659,546 actual shares at the August 27 record date, the newest exact count in the opened documents. Its equity-plan table lists approximately 1.023m shares issuable under outstanding options and awards, versus separate ungranted plan and purchase-plan reserves. Those reserves are not already-issued shares.
A loss-period diluted EPS denominator excludes antidilutive awards, so it is unsuitable as a forward economic share count. Conversely, adding a weighted-period excluded-award figure to all current awards can count the same claim twice. Awards can vest between the dated share register and the next report. We use explicit prospective share budgets rather than claim a precise current fully diluted total.
Our October 2027 budgets are 36m shares in Bear, 35m in Base and 34m in Bull. Lower-priced or slower outcomes leave more room for funding dilution; Bull assumes stronger self-funding and less new issuance. No exercise proceeds are separately credited. The October 20 signed voting-results 8-K does not supply a new operating balance sheet or a fresh financing package.
7. Our initial valuation: FY2028 operating profit, with every bridge visible
We use a forward enterprise-value-to-operating-profit framework. Trailing P/E is unhelpful while GAAP operating earnings are negative and net results include tax benefits. Value per share = [FY2028 revenue × GAAP EBIT margin × selected EV/EBIT multiple + projected October 2027 net financial cash] ÷ prospective diluted shares. EBIT means operating profit after stock compensation and rent.
All following operating paths, multiples, cash budgets and share counts are AHNIM assumptions. The multiples are judgments about growth duration and execution risk, not verified peer medians. Even Bear assumes successful recovery; a failed or delayed ramp can produce values below it. No probability weighting is assigned.
| FY2028 / target-date input | Bear | Base | Bull |
|---|---|---|---|
| Revenue / USDm | 120 | 200 | 300 |
| Gross margin | 40% | 45% | 50% |
| Gross profit / USDm | 48 | 90 | 150 |
| Operating expense / USDm | 36 | 50 | 66 |
| GAAP EBIT / USDm | 12 | 40 | 84 |
| GAAP EBIT margin | 10% | 20% | 28% |
| EV / EBIT | 20× | 30× | 40× |
| Enterprise value / USDm | 240 | 1,200 | 3,360 |
| Target-date net financial cash / USDm | 80 | 100 | 130 |
| Prospective diluted shares / million | 36 | 35 | 34 |
| Calculated value / share | $8.89 | $37.14 | $102.65 |
| Published target / share | $9 | $37 | $103 |
Base requires $200m sales, 42.9% above the FY2027 company-guidance midpoint. Its 45% gross margin creates $90m gross profit; $50m of operating costs leaves $40m EBIT. Revenue, mix, throughput and service economics all have to improve. Bull needs still greater adoption plus stronger gross margins. These outcomes are not established by the existing order announcements.
The 20×/30×/40× EBIT multiples correspond to 2×/6×/11.2× sales at the stated margins. The rising multiples assume longer-lasting growth and more repeat business. The Base 30× already prices substantial success; it is not a cyclical-equipment trough multiple. If purchases stay concentrated and episodic, that premium should fall.
The $80m/$100m/$130m net-cash budgets are projected October 2027 balances, incorporating cash use or generation through that date, not FY2028 closing cash. They imply changes of approximately −$36.4m/−$16.4m/+$13.6m from historical unrestricted May cash. We keep rent in EBIT and therefore exclude operating leases from net financial debt; capitalizing leases would require an accompanying earnings adjustment. Cash is added only once.
At fixed Base revenue, cash and shares, a 15%/20%/25% EBIT margin gives approximately $28.57/$37.14/$45.71 per share. At fixed Base operating profit, 20×/30×/40× gives $25.71/$37.14/$48.57. An extra one million shares reduces the exact Base value to $36.11. These sensitivities expose how much the outcome depends on execution and the selected valuation premium.
8. A supporting DCF makes the cash-conversion burden explicit
Our target-date DCF uses FY2028–FY2032 revenue of $200m/$270m/$340m/$400m/$450m and GAAP EBIT margins of 20%/22%/23%/24%/25%. It applies 25% tax, depreciation of 2% of revenue, capex of 3%, and incremental operating working capital of 15% of revenue growth, beginning from a $140m FY2027 anchor. Stock compensation remains an expense, not an added-back cash benefit.
| Fiscal year — AHNIM | Revenue / USDm | EBIT margin | Full-year FCF / USDm | Post-target FCF / USDm |
|---|---|---|---|---|
| 2028 | 200 | 20% | 19.000 | 12.667 |
| 2029 | 270 | 22% | 31.350 | 31.350 |
| 2030 | 340 | 23% | 44.750 | 44.750 |
| 2031 | 400 | 24% | 59.000 | 59.000 |
| 2032 | 450 | 25% | 72.375 | 72.375 |
With 14% cost of capital and 3% terminal growth, the model gives $511.81m of enterprise value and approximately $17.48 per share, including the same $100m cash and 35m-share bridge. The sensitivity is $14.29 at 16%/2% and $23.05 at 12%/4%. Terminal value equals final-year FCF × (1 + growth) ÷ (cost of capital − growth). Approximately four months of FY2028 precede the target and are already included in projected target-date cash. Assuming even cash accrual, we exclude one-third of FY2028 FCF and discount the remaining two-thirds at 0.667 years; subsequent flows use 1.667, 2.667, 3.667 and 4.667 years. This is an explicit eight-month stub approximation, not an exact fiscal-day model.
The DCF does not independently validate the $37 Base target. The primary multiple case needs a more durable high-growth period, stronger eventual cash economics or lower required returns than this cash path captures. We retain the forward operating-profit scenario as the primary framework, while treating that gap as a reason to demand evidence. If cash conversion fails to improve, we would cut the primary valuation premium.
9. The broker gap: our Base is 71.1% below the accessed mean
The Stock Analysis / S&P Global page reports a $128 one-year mean, $120 median and $110–$175 range. It names six analysts, but five identifiable numeric broker targets across the reviewed pages average $128; the exact target-contributor count is not independently reconciled. Targets were last updated September 10, with an October 8 last-check label. We accessed the page October 22 KST. This older rolling one-year snapshot is only approximately comparable to our October 2027 target, not a synchronized new poll.
| Published AHNIM case | Versus $63.88 reference close | Versus $128 broker mean |
|---|---|---|
| Bear $9 | -85.9% | -93.0% |
| Base $37 | -42.1% | -71.1% |
| Bull $103 | +61.2% | -19.5% |
The absolute Base gap is |37 − 128| ÷ 128 = 71.1%, above our 20% review threshold. Our lower value reflects explicit GAAP operating costs, economic dilution and a selected 30× EBIT multiple. At those same Base margin, cash and share inputs, $128 would require $730m revenue rather than $200m. The $63.88 market reference would require about $356m. Those reverse calculations illustrate our model’s dependence; they do not reconstruct any broker’s valuation.
The Fly’s July 15 reports describe Craig-Hallum raising $68 → $125 and Lake Street raising $56 → $110, both retaining Buy. The former cited AI/hyperscaler growth; the latter cited record bookings and an outlook above its earlier $84m revenue estimate. These are secondary accounts of broker research, not original notes. Both actions precede the October order. No October-order-specific target revision was verified.
A higher Base would need evidence for the changed input: broader production customers, repeat interface demand, higher realized gross margins, controlled engineering and support costs, and collections that fund expansion. A higher consensus alone is not sufficient. A lower Base would follow delayed deliveries, persistent losses, worse working capital or more shares than budgeted.
10. What we will compare with this baseline
| Checkpoint | Evidence that matters next |
|---|---|
| November 2 scheduled results | Separate the 28-day transition from Q1; obtain actual revenue, cash and shares |
| FY2027 company outlook | Compare deliveries and recognized sales with the $130m–$150m range; distinguish GAAP EBIT from adjusted net income |
| August and October orders | Track shipments, acceptance, collections and remaining commitments without adding stale backlog figures |
| Customer expansion | Require independent revenue diversification, not only additional devices at one hyperscaler |
| Operating leverage | Compare realized gross profit, stock compensation and operating costs with our FY2028 bridge |
| Initial $9 / $37 / $103 targets | Retain this dated model; revise identified inputs with new evidence |
The principal risks are customer and purchase-cycle concentration, competition or in-house test alternatives, shipment and acceptance delays, warranty and thermal-performance problems, SiC cyclicality, weaker consumables demand, inventory and collection needs, litigation and further dilution. The asymmetric upside comes from several high-volume programs requiring repeated capacity; the downside comes from paying for that scale before its operating economics are demonstrated.
Related coverage
Our Wolfspeed analysis of SiC manufacturing economics examines the upstream volume and funding cycle relevant to a test-equipment customer base. Our NVIDIA and Cisco optics analysis explains the network-density problem behind another reliability-testing opportunity. These industry links do not establish either company as an identified Aehr customer.
Investment information: This is general research for information and education, not personalized investment advice. Scenario values are uncertain and capital can be lost. We use AI to assist research organization, calculations and editorial illustration; original documents and dated corrections govern the analysis. See our research and correction policy.
Sources and original documents
[1] FY2026 Form 10-K — Filed July 27, 2026; year ended May 29.
https://www.sec.gov/Archives/edgar/data/1040470/000165495426006919/aehr_10k.htm
[2] Q4/FY2026 results and GAAP/non-GAAP reconciliation — July 14, 2026.
https://www.sec.gov/Archives/edgar/data/1040470/000165495426006655/aehr_ex991.htm
[3] FOX wafer-level AI production order — August 12, 2026.
https://www.aehr.com/2026/08/aehr-receives-22-million-follow-on-order-for-ai-processor-wafer-level-burn-in-systems/
[4] Sonoma follow-on production orders — October 7, 2026.
https://www.aehr.com/2026/10/aehr-receives-follow-on-sonoma-production-orders-for-lead-hyperscale-customer-s-next-generation-ai-processor/
[5] Silicon-photonics production order — June 17, 2026.
https://www.aehr.com/2026/06/aehr-receives-follow-on-order-from-major-silicon-photonics-customer-for-fully-automated-wafer-level-burn-in-system-for-hyperscale-data-center-optical-interconnect/
[6] Fiscal-year change, Form 8-K Item 5.03 — April 7, 2026.
https://www.sec.gov/Archives/edgar/data/1040470/000165495426003310/aehr_8k.htm
[7] Official Q1 FY2027 reporting calendar — October 8, 2026.
https://www.aehr.com/2026/10/aehr-test-systems-to-announce-first-quarter-fiscal-2027-financial-results-on-november-2-2026/
[8] Definitive proxy and equity-plan information — 2026 proxy; share-count record date August 27.
https://www.sec.gov/Archives/edgar/data/1040470/000165495426008222/aehr_def14a.htm
[9] Annual-meeting voting results, Form 8-K — Signed October 20, 2026; meeting October 19.
https://www.sec.gov/Archives/edgar/data/1040470/000165495426009056/aehr_8k.htm
[10] AEM AMPS-BI announcement — July 9, 2024.
https://aem.com.sg/news/aem-introduces-the-new-generation-of-automated-burn-in-systems-for-ai-and-high-performance-compute-chip-testing/
[11] Micro Control AI/HPC burn-in offering — Undated page, accessed October 22, 2026.
https://www.microcontrol.com/application/ai/
[12] Stock Analysis: AEHR market quote — October 21 regular close; accessed October 22 KST.
https://stockanalysis.com/stocks/aehr/
[13] Stock Analysis / S&P Global: forecast and analyst history — Targets updated September 10; last-check October 8; accessed October 22 KST.
https://stockanalysis.com/stocks/aehr/forecast/
[14] The Fly via TipRanks: Craig-Hallum target increase — July 15, 2026; original broker note not accessed.
https://www.tipranks.com/news/the-fly/aehr-test-systems-price-target-raised-to-125-from-68-at-craig-hallum-thefly-news
[15] The Fly via TipRanks: Lake Street target increase — July 15, 2026; original broker note not accessed.
https://www.tipranks.com/news/the-fly/aehr-test-systems-price-target-raised-to-110-from-56-at-lake-street-thefly-news
Leave a Reply