Silicon Carbide Gold Rush – Who is Selling the Drilling “Equipment”?
In our opinion, there’s an interesting company that stands to benefit in the silicon carbide space on Kato Road in Fremont, California and it is not Tesla. We believe Aehr Test Systems (symbol:AEHR) presents the ability to become an owner of a relatively unknown yet profitable company just as it has turned to profits. When companies can turn on the profit switch, in our experience there is the potential for profits to move higher as they grow from a small base. Aehr sits in an industry with incredible tailwinds, has growing revenues, and has a product that management claims is eighteen times (18x) more capable in some respects than their competitors.
Important Background
Due to the rapid success of the electric vehicle (EV) industry, the need for silicon carbide (SiC) computer chips (aka “dies” or “devices”) is growing exponentially. SiC is able to operate in a high voltage, high temperature environment of an EV’s traction inverter (especially rear vs front), on-board charger, DC-DC inverter, and power management system. Silicon carbide inverter can drive a range benefit of 5-20% in EVs according to Wolfspeed CEO Gregg Lowe. Yet the 97-99% yield on manufacturing SiC at the wafer (or “substrate”) level is nowhere close to the yield of 99.99%+ for memory chips.
To the uninitiated, a failure rate of 1% on a single die may not sound bad. A standard traction inverter has 48 SiC devices on one computer board or “package.” For the traction inverter to properly function, not one of those dies can go bad. Additionally, the failure of the die can happen 6 months down the road because of the tremendous stresses of operating in high temps & voltages. This failure means the car stops working (perhaps in a dangerous situation) and expensive repairs.
To solve this “infant (chip) mortality” issue, “burn-in” testing is done to subject the silicon carbide dies to a year’s worth of stresses thereby aging the devices past their growing pains. This happens over the course of 10 hours to a few days. Once a die has successfully made it through burn-in, it is largely ready but is still tested by the auto component supplier at the package level. If you wait to test at the package level with 48 dies, you may be throwing away 47 good dies and one bad die.
The recipe for burn-in is temperature, voltage, and time. Traditionally the temperature and voltage are at a higher level than normal operating parameters. The objective is to apply heat and voltage to cause the bad parts to fail in the process. In EVs, this means having the device fail on the wafer rather than failing in a moving vehicle or in a vehicle under drivetrain warranty.
Chipmakers may adjust this recipe to increase temperature and voltage in the hopes that this additional stress will result in lower burn-in times (cause the failures earlier in the burn-in process). Lower burn-in times would result in higher throughput and capacity (lower cost for the customer). A risk of increasing the voltage is this additional stress may damage good parts. The intent of burn-in is to weed out infant mortality (bad parts). You do not want to damage good parts.
In terms of testing, the “functional test” process is separate from burn-in. The example I use is a light bulb (simple example). You flip the switch and the light bulb goes on (“functional test”). However, just because the light bulb goes on when you flip the switch does not mean it will stay on after the 100th or 1000th switch or after one month or one year. So before burn-in is performed there is a functional test on the wafer to tell good die from bad die. You then perform burn-in to stress the wafer. You then perform a functional test after burn-in to identify the bad parts after the burn-in process.
This potential waste makes it incredibly important to do Wafer Level Burn-In and Testing (WLBIT). By testing at a wafer level, the chipmaker can have extreme confidence which dies to put on a package. This saves a tremendous amount of time and resources as you can imagine. This is where companies like Aehr Test Systems come in.
Proprietary Solution
High temperature “ovens” and high-voltage electrical probes are nothing new to preparing dies for commercial sale. These probes make thousands of contacts on the wafer during burn-in. This has been happening in silicon for decades. Aehr uses patented “probe-less” wafer level burn-in and testing system that can test eighteen (18x) wafers at a time. The probeless, contactless nature of Aehr’s FOX-XP system allows it to test 18 wafers simultaneously. We believe Aehr’s representations that their next closest competitor in WLBIT can only provide a solution to test one wafer at a time in a similar footprint without breaking Aehr’s patents.
With burn-in taking half a day for a wafer, the “old way” means one system allows WLBIT of 2 wafers per day or roughly 700 per year. Aehr’s system allows for a factor of 18 improvement… 36 wafers per day or 12,600 per year. Note: please don’t hold these estimations to be fact, this is for illustration only, do your own math. This gives Aehr a strong competitive “moat” allowing them to price their product more attractively per unit of throughput than competitors.
The defining factors of monopolistic power in our opinion is being able to deliver a product to your customers at a superior unit value per dollar than your competitors. While there is technically more than one company operating in the WLBIT niche—and by dictionary definition Aehr is not a monopoly—we believe their technology represents a unique compelling solution for silicon carbide chip makers. Effectively Aehr can provide more value at lower cost.
High Level Financials & Guidance
Aehr is a small company whose business we believe hit an inflection point earlier this year. For their fiscal year ending May 31, 2022, Aehr had net sales of $50.8 million vs $22.3 million two years prior (mostly before the health crisis began). Last fiscal year they made $9.5 million profit, vs a net loss of $2.8 million two years ago. Before the crisis, Intel directly and Apple suppliers validated, purchased, and installed Aehr’s systems primarily for WLBIT of 3D sensing chips. ON Semiconductor Corporation is the largest customer for Aehr now, more about them in a bit.
Aehr’s future business is hard to predict. Because of (1) how relatively new the FOX-XP product is, (2) how small Aehr’s business is today, and (3) the fact that an automated FOX-XP system with 18 testing “blades” costs $4-4.5 million. If all they sold was new systems and the consumable for those systems it would have only taken 12 systems to get to $50 million in net sales. Said another way, if they sold one less FOX-XP last year, their revenues would have otherwise been lower by 8-9%. This binary outcome of their sales process will make this phase of their growth particularly lumpy. This will evolve over time but is a key risk.
In our opinion, CFO Ken Spink and CEO Gayn Erickson have been conservative in their full year guidance over time. Last July 15, 2021 they guided for revenue to exceed $28 million which would have been 70% growth over the previous year. Instead, as stated above Aehr did $51 million in net sales. They did revise the guidance on March 31, 2022 from previously stated $30 million to $50 million.
We believe Aehr is still being a bit conservative based on our checks of the largest and most serious silicon carbide device makers. ON Semi plans to now triple their projected 2022 ($500 million revenue) to a $1 billion run rate exiting 2023; this represents 40%+ growth. STMicroelectronics plans to grow from $700 million to $1 billion in a year; 43% growth. Leapfrogged by ON Semi, Infineon plans to grow from $200 million in SiC revs to $360 million over their next fiscal year or 80% growth and grow to $1 billion by mid 2020s. From $360 million to $1 billion in the following 3 years would be about a 42% CAGR.
One should think of the SiC chipmakers as the gold prospectors and Aehr’s FOX-XP system as going from shovels to drills. If the SiC makers do not grow their capacity or change their chip designs, Aehr’s revenue would fall off a cliff. Chipmakers could have capacity growth of 0% and AEHR net sales shrink by 70%. This is another key risk. But it also is a source of upside opportunity as the chipmakers are selling to the EV makers who grew sales by 100% last year.
This would be a risk if numerous SiC were NOT growing capacity, or if they were growing linearly. The opposite is the case. ON Semi, STMicro, Infineon, and Wolfspeed are all building new fabs (aka factories) that are primarily dedicated to silicon carbide. Infineon’s 2024 $2 billion factory in Malaysia will focus on SiC and gallium nitride (GaN) which FOX-XP systems can WLBIT as well.
Our guess is 40% growth may be a minimum and maybe all these players are being conservative. But not all Aehr’s business is with the FOX-XP system, and not all the FOX-XP systems are used for SiC WLBIT. Systems are used for silicon photonics (networking), gallium nitride (solar), and possibly regular silicon in the future. On Aehr’s July 19, 2022 profits call, Aehr management guided for $60 to $70 million in net sales for the year ending May 31, 2023. This would represent 20-40% growth. How might this translate to profits?
Direct Margin
On the latest call, we asked CEO Gayn Erickson directly about Aehr’s incremental gross profit margin… Here’s the exchange from the transcript (edited for clarity).
Bradford Ferguson: …Revenues versus last quarter grew 5 million. But your COGS only grew a million. So even if you adjust for that inventory write down that happened (last quarter), it seemed like your COGS only grew by 2 million. So to me that says you have a potentially a 60% gross margin as you scale bigger and bigger?Gayn Erickson: You are within a point. You are right. You're dead on. Yes, are what we call our direct margin, which is basically our total costs, including warranty, et cetera, are about $0.40 on the dollar. And our margins are actually improving from last year to this year in direct margins. As we go to 200 millimeter WaferPAKs, our margins are actually improving.
Direct margin is how much the company would make all-else-equal for each additional dollar of revenue. What Gayn is saying is that for each extra dollar, they believe their direct costs are 40 cents. We expect some additional indirect costs to come with Aehr’s projected growth. Over the past year, stock-based compensation grew by $1.8 million vs $34.2 million growth in revenues (5.3% or 5.3 cents). Also, growth in SG&A and R&D expense was $5.6 million vs $34.2 million growth in revenues. (16.4% or 16.4 cents).
Inventory write-downs were excessive in the past year as Aehr switches its focus to FOX-XP, this cost $900,000 extra than what may be usual. Aehr also benefitted last year from forgiveness of a PPP loan to the tune of $1.7 million. This leaves us with an adjusted 2022 earnings of $8.65 million without factoring in taxes.
We believe $70 million net sales for fiscal year 2023 is more realistic than $60 million. When we look at a potential $70 million for this year, additional earnings before tax (EBT) SBC-adjusted profits could be $7.3 million (factoring in higher SBC & SG&A costs) for a total SBC-adjusted EBT of $16.0 million (counting SBC as a cost). If we are owners of a business, giving shares to employees dilutes us or is “an expense” of sorts, in our opinion. Do your own research. You can calculate how you like using Aehr’s most recent 8-K SEC filing.
Then perhaps you take 20% out for taxes long-term and you arrive at $12.8 million in net income. $12.8 in income would be 48% in profits growth vs our $8.65 million adjusted number.
On diluted shares outstanding of 28,568,000 that comes out to $0.45 per share. If you put a 40 multiple on that, it comes out to $18/share for 2023.
If you assume that Aehr revenue only grows 40% the subsequent year, using heavy extrapolation, that might mean an extra $28 million in revenue, an added $10.7 million in EBT SBC-adjusted, a total of $26.7 million EBT and $21.3 million net income after 20% tax rate. In this scenario, earnings would have grown by 66% for 2024. Applying a 40 multiple to net income would be $29.95/share for 2024.
One more year of 40% revenue growth may mean an extra $39.2 million in revenue, an added $15.0 million in EBT SBC-adjusted, a total of $41.7 million EBT and $33.4 million after taxes. In this scenario, earnings would have grown by 56% for 2025. Applying a 40 multiple to net income would be $46.73/share for 2025. Overall, this hypothetical would be a conservative base case in our opinion. We will share the bull argument after this brief detour.
Aehr’s Large Customer
Recently ON Semi has adopted Aehr primarily for testing of silicon carbide. In our research ON Semi seems to be a serious player in the SiC space, focusing on quality throughout the production steps and higher margin products. On its most recent profits call, Aehr CEO Gayn Erickson admitted that it’s largest customer for the past year represented 82% of its revenues. This represents a significant financial risk if ON Semi were to change its plans.
We don’t think they will as ON Semi has gone from a 6 or 7th ranked player in silicon carbide to a top 2 player today. ON Semi is focusing its capacity expansion largely on the SiC space. Many SiC players lead their investor presentations talking about SiC even if today a majority of their revenues come from regular silicon products, SiC is where the growth is. Don’t take our word for it.
"We are continuing to make progress on our silicon carbide growth and remain on track to more than double our silicon carbide revenue in 2022 as we ramp shipments to customers who have signed long-term supply agreements with us. At this pace, exiting 2023, ON Semi will be on a $1 billion run rate for silicon carbide revenue.” ON Semi CEO Hassane El-Khoury. May 2, 2022ON Semi is putting its money where its mouth is: “As indicated previously, we are directing a significant portion of our capital expenditures towards enabling our 300-millimeter capabilities at East Fishkill fab and the expansion of silicon carbide capacity. This increase is in line with the higher capital intensity in the near term, as mentioned at our Analyst Day.”
Though we do not know when, we don’t believe that ON Semi remains such a large percentage of Aehr’s revenues even over the next year. We believe that Aehr’s product is so compelling: the savings of time, materials, labor, and floorspace add up over time. Most importantly, FOX-XP 18 WLBIT allows for SiC makers to quickly scale production.
Aehr Bull Case
Both STMicroelectronics and Infineon are past customers of Aehr. Wolfspeed is not on Aehr’s list of customers, yet Wolfspeed has 20 years’ experience in silicon carbide and likely will understand the value. Aehr revealed during their call that they delivered over $10 million of revenue in the single month of May 2022. If they had the orders, they are confident they could sustain a $10 million per month rate for a full year. ON Semi essentially did $40 million with Aehr for fiscal 2022, a large series of orders from STMicro could rival $40 million, or a combination of smaller orders from smaller players could do the same.
SiC is only projected to grow by 40% while the EV space is growing faster. If the EV use of SiC becomes a bigger percentage of SiC demand, then SiC capacity will need to grow faster than 40% to meet that demand. 40% SiC growth might also be sandbagging from the chipmakers.
Additional upside could also come from silicon photonics (SiPh) makers. SiPh is used in big data centers as a replacement for fiber optic cables. It enables greater bandwidth and therefore speed. According to Aehr, SiPh was implemented in data centers pre-pandemic, but capacity among the SiPh has not increased much since then. Should this area re-accelerate, meaningful customers could emerge for Aehr’s FOX-XP testing solution.
We have also counted research & development (R&D) and selling, general & administrative (SG&A) costs as expenses where the money is just “gone.” Yet R&D, when spent well, can be an important investment and driver of future profits. For example, Aehr spent some resources to make its product be able to serve more use cases and to improve usefulness of the Aligner. It’s also possible that R&D spend will moderate, more clarity is needed here. Also if these costs are constrained more as Aehr grows, it could add to short-term financials.
With how long Aehr’s sales cycle is (sometimes taking over a year), much of its past sales expenses could be viewed as an investment for closing future orders and thus drive profits. The traditional accounting model is to count all of R&D and sales expenses as gone and that is likely not the full story.
Aehr Bear Case
We already mentioned the risk with their large single customer, ON Semi. The binary nature of a single system delivered being 8% of revenues is also a risk. Sales may not come through. Potential customers may not see the value or they may not have the budget for the upfront cost. Expenses at Aehr could also run higher than expected.
There is a depth of information to the silicon carbide industry and we have not uncovered everything nor thought through all possible probabilities (that’s what AI may be for). There are materials and exhibits from Aehr that we have not fully digested—primarily the impact of outstanding stock options for insiders.
Also note, as a small-cap growth company, Aehr's stock is a member of several "total stock market" and "extended market" ETFs that have been panic sold by investors fearing higher inflation. As such, we believe these ETFs were forced to sell AEHR stock as investors redeemed their ETF shares. When ETFs sell the stocks held in their basket, they do not care about the price they are selling the stocks. They are not price sensitive. It was sold all the way down to becoming a micro-cap company (under $250 million).
AEHR stock can be thinly traded. Prior to the writing of the article, there were days when less than $2.8 million in stock exchanged hands in a single day. Buyer and seller beware and in a name like this it may not be wise to use market orders. Despite being a small-cap stock, we’ve seen evidence of both highspeed and algorithmic trading.
Conclusion
This report is to serve as an introduction to Aehr Test Systems. As of 2019, there were more than 40,000 stocks worldwide and we believe Aehr suffers from being a small company with a stock that has a long, somewhat uninspiring history. We’re putting this out there to show how one can use extrapolation to begin to value a company’s stock, not to serve as “the way.” Our calculations may be done in a way that few agree with, so do your own math! We are including a list of source material below (including for the SiC chipmakers) and urge you to go to the source; see and think for yourself.
Buyer or seller beware! We’ve not audited Aehr Test Systems nor yet visited their facilities. We are at the relative beginning of our research into the SiC, SiPh, and GaN spaces. This may be the last stock research report we post, so do not count on us for updates.
Given the risks in the business, we think it is fair to have an eye on the conservative case presented, but to know there could be even greater upside if Aehr gains new large customers and if those customers grow more quickly than they are saying. Silicon carbide chipmaking is a competitive business and we believe that Aehr will benefit materially as the chipmakers make billions in investments. This is all provided that their solution is chosen.
All this said our targets with all the knowledge we have today are $18/share for summer 2023 (through May 31, 2023), $30 for summer 2024, and $47 for summer 2025. We are initiating a rating of “overweight.” Our opinions are subject to change.
Bradford Ferguson, CFA
Tony Hutton
Sources
AEHR profits call transcript 7/19/22
AEHR investor presentation slides 12/7/21
AEHR SEC filings
AEHR CEO Gayn Erickson YouTube presentation 11/6/21
ON Semi earnings call transcript 5/2/22
ON Semi earnings presentation 5/2/22
ON Semi investor presentation slides
STMicroelectronics earnings call transcript 4/27/22
STMicroelectronics investor presentation slides
STMicroelectronics conference presentation
Infineon earnings call transcript 5/9/22
Infineon investor presentation slides
Infineon conference presentation
As of July 21, 2022, Clients and employees of our firm Halter Ferguson Financial own AEHR stock and/or options and thereby stand to materially benefit from a rise in the share price. Past performance is no assurance of future results. Halter Ferguson Financial, Inc. (“Halter Ferguson Financial”) is a registered investment adviser with its principal place of business in the State of Indiana. A complete list of all recommendations will be provided if requested for the preceding period of not less than one year. It should not be assumed that recommendations made in the future will be profitable or will equal the performance of the securities in this list. Opinions expressed are those of Halter Ferguson Financial, Inc. and are subject to change, not guaranteed and should not be considered recommendations to buy or sell any security.