Long AEHR Test Systems (NASDAQ: AEHR)

15-Dec-22

The Elevator Pitch

AEHR provides test systems for burning-in semiconductor devices in wafer level, singulated die, and package part form. Through a project with US government agency DARPA back in the 1990s to early 2000s, AEHR developed superior hardware and software capabilities in wafer-level test and burn in. However, this technological capability which they possessed was way ahead of their time hence the lack of commercialization opportunities. Fast forward two decades later, the time is ripe for AEHR to commercialize their capabilities in the SiC (silicon carbide) and SiPh (silicon photonics) markets backed by demand for EVs and I/O performance bottlenecks respectively. Today, AEHR is the only player in the semiconductor equipment space that can offer high volume wafer-level test and burn-in for SiC chips with no competitors in sight. AEHR’s revenue broke out +206% in FY22 and has now reached a permanent inflection point in its long-term top and bottom-line size driven by the existing commercialization opportunities.

This is a small cap stock with 0 eyeballs from the street. While stock has reacted and factored in FY23’s all-time high guidance, I believe street has yet to price in the magnitude of ramp-up from AEHR’s SiC customers (ON Semi, Infineon, ST Micro, and Wolfspeed), and Si-Photonics as a second growth engine.

I value AEHR on equal-blended 2 year forward 6.0x P/S and 35.0x P/E and arrive at a target price of $37.36, +61% upside from 15 th Dec 2022’s closing price of $23.21. I see two risks from this GARP play – i) A faster than expected materialization of consumables’ growth and revenue contribution surpassing that of systems, and ii) Lack of catalyst in driving Si-Photonics as a second growth engine. This translates into a -20% upside, which brings me to an overall risk-to-reward of 1:3.

The SiC Thesis

AEHR has reached a permanent inflection point in its long-term top and bottom-line size driven by commercialization opportunities of its technological capabilities that was previously “way ahead of its time”. Demand for SiC is the first growth engine driving AEHR.

Historically, high barriers to entry in a niche and small TAM and a “way ahead of its time” technological capabilities meant lack of commercialization opportunities. This explains the under US$30mn per year with low single digit to negative growth revenue profile of AEHR. High operating leverage nature of AEHR’s business greatly impacted profitability where there were more unprofitable (19) than profitable (9) years since AEHR’s listing.

However, things have started changing. AEHR’s revenue exploded by +206% from US$16.6mn in FY21 to US$50.8mn in FY22. Net profit came in at US$9.45mn, positive for the first time in 3 years, the highest in 14 years. This is not a one-off event, but rather a permanent inflection point in how AEHR’s business will scale in the long run.

Revenue breakout in FY22A was not a one-time off event, but rather a permanent inflection point driven by commercialization opportunities in SiC

EV is built upon a powertrain which requires ~2 traction inverter. Each traction inverter requires ~45-50 SiC MOSFETs. To manufacture SiC MOSFETs, you require wafer-level testing, which AEHR is able to provide at the highest capacity while at the lowest cost

The first driver of change for AEHR today is the demand for WLTBI of Silicon Carbide (SiC) devices which are used directly in the electric vehicle (EV) market today. SiC-based devices such as metal oxide semiconductor field effect transistors (MOSFETs) are extremely efficient, rugged, and reliable semiconductors used in power conversion to charge the batteries and EVs, and in the engine controller traction inverters that drive the electric engines in EVs. The key differentiators of SiC over Si-based insulated gate bipolar transistors (IGBTs) include SiC’s higher system-level efficiency due to greater power density, lower power loss, higher operating efficiency, and increased temperature operation. This effectively translates into higher driving range on a single charge, smaller battery sizes for traction inverters, and faster charging time for onboard chargers. SiC MOSFETs are tested to ensure they meet technical performance and specifications before they're shipped to customers.

The bottleneck and technical challenges for any SiC MOSFET suppliers today is the overall state of SiC MOSFET yield, reliability, and costs where it is currently 30 years behind Si. Importantly, yield loss is driving ~30% of production cost today. This is due to the properties of SiC where it has high defect density and lower purity over Si. Meaning to say, the early life failure rate of SiC MOSFET is much too high for mission-critical applications such as traction inverters or even the onboard chargers. The high early life failure translates into low yield, hence the high yield loss.

KGD is the fundamental element needed in producing SiC MOSFETs. AEHR FOX systems’ WLTBI solution is the fastest and cheapest way of identifying KGD.

This is where AEHR comes into the picture. AEHR’s ability to conduct WLTBI allows for production of known good die (KGD), a fundamental element essential for multi-chip modules such as three-dimensional integrated circuits (3D ICs). In this case, AEHR’s FOX systems can produce high volume KGD for SiC MOSFETs. There are 4 major SiC suppliers in the industry today – ON Semiconductor, Wolfspeed, ST Microelectronics, and Infineon. 2 out of the 4 players (ON Semiconductor and another undisclosed), are now ordering AEHR’s FOX Systems on a high-volume level. The remaining 2 players are currently on an evaluation level with AEHR.

Why now?

I expect AEHR to secure the remaining 2 players for high-volume production by end FY23, and also for AEHR to meet upper range of their FY guidance (US$70mn). From Jul-22 to Nov-22, AEHR announced 4 FOX-XP, 2 FOX-NP, and US$8.4mn of consumables for delivery by FY3Q23. This means that current visible revenue for FY23E comes to ~US$44mn and that AEHR will only need another US$26mn of orders to meet upper range of guidance. Beyond FY23E, I expect AEHR to easily sustain >US$120mn revenue for a minimum of 2-3 years due to the aggressive ramp up from the 4 major players.

Our bullishness on AEHR stems from the ramp up needed to produce ~31mn EVs by FY30E. Canaccord Genuity estimates wafer capacity to increase 25x from 150k pieces of 6-inch wafers in 2022 to 4mn pieces of 6-inch wafers in 2030. They also estimated another 4mn pieces of 6-inch wafers required in addressing other end market applications, specifically for industrial and solar power conversion. Based on Gayn Erickson’s (AEHR CEO) estimates, every 1mn of EVs requires 8 units of their FOX-XP systems. Assuming 24hrs of test time per wafer (which can actually go up to 48hrs) and an average ASP of ~$4mn per system, revenue will need to sustain a steady run rate of at least US$120mn per FY. Current calculations are on FOX systems alone and exclude the consumables that are required whenever new SiC chip design comes along.

The SiPh Thesis

The 5G and data centre space is quickly approaching practical limits of electrical input/output (I/O) performance. As bandwidth demand for compute keeps increasing, electrical I/O is not scaling to keep pace, resulting in an “I/O power wall” that limits available power for compute operations. Meaning to say, the world has reached a point where the chip can process information faster than electrical signal can get in and out of it. High production cost of fibre-optic transceivers is another ongoing industry challenge that has yet to be addressed.

Si-Photonics transceiver is a solution in addressing the I/O power wall due to the much higher data rate they can offer over traditional optical technologies. For example, it can support data rates of up to 100 Gbps, which is 10 times faster than the best fibre-optic transceivers offered in the market today. The properties of SiPhotonics using less energy and heat in generating light also allows it to consume less power than traditional optical technologies. All in, Si-Photonics allows for much higher photonic complexity than discrete designs, delivering improved performance, better power efficiency, and better cost at scale. The superior technical properties and commercial benefits over conventional methods is driving demand for Si-Photonics today.

AEHR comes into the picture again on a WLTBI aspect. The lasers used in Si-Photonics transceivers requires an aging / stabilization process where precise and calibrated currents are applied through every laser device at elevated temperatures to wring out process variations and to stabilize the output power of each device. This process takes many hours to even days to do and is needed on 100% of devices. AEHR’s FOX-XP system is capable of not only having all the power supplies and the capability to 100% discretely determine and test every single laser that are measured in thousands per wafer in one shot, but also can deal with the amount of power to the devices and removal. The ultimate value proposition from AEHR is a cost-effective enabler for Si-Photonics devices in ensuring it is reliable for application use.

Si-Photonics for 5G and data centre was a rising theme for AEHR back in 2019. Intel contributed to more than 20% of AEHR’s sales from FY19 to FY21. Outbreak of Covid-19 and deteriorating macro environment thereafter in CY22 restricted the Si-Photonics story from properly taking off. However, commentary and roadmaps across key players within the space such as Intel, Marvell, and Broadcom suggest Si-Photonics is poised for acceleration sometime in CY23.

I expect Intel to resume their capex spending for AEHR’s FOX systems in 2H23 and estimate ~US$24mn of revenue contribution to AEHR in FY24. Considering there is 0 coverage from sellside, I believe street has yet to factor the Si-Photonics thesis into stock price today. Consequently, I expect a meaningful rerate on P/S multiple should this thesis play out.

How to think about valuation today

I value AEHR on equal-blended 2 year forward 6.0x P/S and 35.0x P/E and arrive at a target price of $37.36, +61% upside from 15 th Dec 2022’s closing price of $23.21. I believe AEHR rightfully commands a valuation premium over peers in the semiconductor equipment tester space for the WLTBI moat that they possess, the lack of competitors in taking share from them, and the high visibility of both theses playing out.

Catalysts

1. The stock today moves primarily on announcement of new orders. Announcement of new FOX-XP orders from major SiC customers in the coming months will provide further visibility on path towards US$70mn by end of FY23. Lead time for delivery of systems is ~5 months.

2. Ability to qualify the remaining 2 major SiC customers from evaluation stage to high volume production stage by end of FY2Q23 implies possibility of delivering orders for all 4 major SiC customers in FY4Q23. This would see significant boost on Q4 revenue. Typical sales cycle for evaluation stage is anywhere from 9 months to over a year. However, ON Semiconductor accelerated the process to under 7 months. Hence, there is a real possibility of delivering orders for all 4 major SiC customers in FY4Q23.

3. FY24E topline guidance will set the tone for next 2-3 years of revenue run rate.

4. Announcement of FOX-XP orders from a major Si-Photonics customer (Intel) /Intel being a top 5 customer with more than 10% revenue contribution in FY23E / Announcement of new Si-Photonics customers entering the evaluation and qualification stage in the upcoming quarters.

What if I turn out wrong

There will come a point in time where consumables’ growth and revenue contribution surpass that of systems. This implies that SiC customers are done ramping up on FOX-XP systems and AEHR would likely see a significant slowdown on revenue growth. Effects of operating leverage would also come into play. All in, we would see material impact on both top and bottom-line. This risk is more probable of materializing in FY24E vs. FY23E since we are only at the start of the ramp up. This risk translates into -US$40mn on topline and 3% erosion on gross margin.

The roll out of Tesla Model 3 in 2018 where Elon Musk announced the integration of SiC MOSFET modules into his new EV line was the catalyst that drove all OEMs EV startups to adopt SiC chips. That marked the start of the SiC story taking off. For Si-Photonics to be AEHR’s second growth engine, catalysts for investment thesis 2 will need to materialize. The lack of which will invalidate investment thesis 2. This translates into -US$24mn on topline based on our assumptions.

Risk to Reward

Should the above 2 risks play out, I would see a -20% upside vs. a base case of +61% upside from 15th Dec 2022’s closing price of $23.21. This gives an overall risk-to-reward of 1 : 3.

Wafer-Level Test & Burn-In (WLTBI)

Semiconductor manufacturing is a complex process. Defects or weaknesses that may result in the failure of a semiconductor device may occur at any process step. Failures may occur immediately or at any time during the operating life of the device, sometimes after several months of normal use. Semiconductor manufacturers hence rely on testing and reliability screening to identify and eliminate defects that occur during the manufacturing process

WLTBI refers to the process of subjecting semiconductor devices to electrical testing and burn-in in wafer form. Burn-in is a temperature/bias reliability stress test used in detecting and screening out potential early life failures.

WLTBI usually employs a wafer prober to supply the necessary electrical excitation to all the die on the wafer through hundreds or thousands of ultrathin probing needles that land on the bond pads, balls, or bumps on the die. The required die temperature elevation, on the other hand, is achieved by the wafer prober through a built-in hot plate that heats up the wafer to the correct junction temperature.

The ideal semiconductor manufacturing scenario is to come up with a process that does everything at wafer level. An integrated wafer-level packaging, wafer-level electrical testing, and wafer-level burn-in will streamline the over-all semiconductor manufacturing process to a large degree, resulting in great cost savings and much shorter cycle times.

Industry Barriers to Entry/Disrupt

The challenge in any wafer-level testing and burn-in process is being able to use existing wafer probing technology to contact all the operation-essential pads of all the die on the wafer at the same time. This is referred to as full-wafer or whole-wafer contact technology. The ability to do so will allow the burn-in process to be conducted to the entire wafer in one operation.

In an article by Dan Inbar and Mark Murin of M-Systems (source: Semiconductor International, 8/1/2004), the formidability of achieving whole-wafer contact with today's wafers was explained using a simple example: if a typical wafer has 500 die, with each die containing 40 functional pads, then 20,000 probing points are needed to properly activate all of these die on the wafer during burn-in. Cramming all of these probe needles onto a single 6" wafer at the same time without allowing any of them to come into contact is indeed challenging.

Current industry standards can only conduct Testing and Burn-In at package-level. Failing at package-level means all costs from assembly process would have gone to waste.

Company Overview

AEHR provides test systems for burning-in semiconductor devices in wafer level, singulated die, and package part form. The company’s main product line is its FOX systems which allows for wafer-level burn-in and tests. There are two main product line within the FOX systems – i) FOX-XP and ii) FOX-NP.

FOX-NP system can test and burn-in 1 wafer at a time. Customers usually order this system for initial production qualification and configuration purposes. ASP range from $400k to ~$1.5mn. FOX-XP system can test and burn-in up to 18 wafers at a time. Customers usually order this system after passing the initial production qualification and configuration stage. ASP range from $1mn to $2.5mn. Key components within the FOX system are AEHR’s consumables - WaferPak Contactor (wafer-level probe card) and DiePak Carrier (sockets). ASP range from $60k to $150k per piece.

The small TAM (Under US$1bn) reflects the nature of AEHR’s top line profile where revenue is typically under US$30mn per year with positive single-digit to negative growth. The high operating leverage nature of this sector resulted in a volatile bottom line profile for AEHR with net profit swinging wildly year-on-year. Overall, there were more unprofitable (19) than profitable (9) years since its listing.

Competitive Advantage

AEHR has a technological moat that is difficult to replicate. This is backed by AEHR’s patented hardware and software architecture that are years ahead of any competitors in the test and burn-in market today. AEHR’s source of moat came from a particular event 28 years ago. In 1994, AEHR entered into a cost-sharing agreement with US government agency Defense Advanced Research Projects Agency (DARPA) in developing WLTBI systems for the production Known Good Die (KGD) that is essential in critical mission applications i.e., satellites and warfare machineries. The success of the DAPRA project moved the needle in positioning AEHR as a significant player in offering WLTBI solutions for mission critical applications. While there are some SME scale players in the WLTBI space today, none compares to what AEHR can offer. On a per wafer level, AEHR is 6.5x cheaper and requires 35.71x lesser footprint. What this effectively means is that AEHR is the only company in the world that can offer WLTBI solutions at the highest volume while at the lowest price. The comparison table below quantifies it all.

Competitive Landscape

WLTBI is a niche space with a relatively small TAM under $1.5bn in the last two decades prior to 2018. The high barriers to entry in WLTBI space and the lack of commercialization opportunities meant that this was an unattractive space for market leaders such as Teradyne and Advantest to venture into. There were a few WLTBI players in the earlier years. However, through economic cycles and downturns, many of the competitors went away. AEHR’s ability to survive through the last two decades of cycles ultimately positioned them as one of if not the only listed WLTBI pureplay in the industry today.