MRAM - Everspin Technologies: The Rally May Not Have Much Further To Go

2023-10-16 02:57:49 ET

Summary

• The stock has rallied, especially in recent months, but MRAM may have a hard time going higher for a couple of reasons.
• MRAM could stand to benefit from AI, which has assisted the stock’s rise, but there are no guarantees this will happen.
• MRAM is arguably in the lead, but there are other emerging types of non-volatile memory, which could eventually displace MRAM.
• The decision whether to go for long MRAM comes down to if the technology is likely to emerge a winner in the race for persistent memory.

Everspin Technologies ( MRAM ), the leading provider of magnetoresistive random-access memory or MRAM, has been on a tear, especially in the last half year or so. The stock has soared higher thanks to a number of reasons, including better-than-expected earnings and upbeat guidance, but also because of the perception that MRAM could benefit from the rise of artificial intelligence or AI. However, the stock may be running out of steam. Why will be covered next.

Why MRAM is worth a look

MRAM is a type of non-volatile random-access memory, which means it has the ability to retain data even if it loses power. What’s so appealing about a non-volatile memory like MRAM is that it essentially seeks to combine the strengths of the two most common types of memory chips, dynamic random-access memory or DRAM and flash memory, into one. This could have a number of benefits, at least in theory.

For instance, MRAM will not lose its contents once power is turned off like DRAM, similar to flash memory, yet without incurring the performance penalty of flash memory compared to DRAM. The combining of separate chips and their functions into one can save space and increase efficiency. On the other hand, the disadvantages of MRAM include its manufacturing complexity and high cost. MRAM has a less dense structure, although STT-MRAM improves on this drawback. MRAM could also be less reliable, especially in an environment where there are strong magnetic fields.

However, there are some applications where the strengths of MRAM could outweigh its weaknesses. One of the more intriguing possibilities is AI. For instance, in HPC applications like AI, the main bottleneck is increasingly the amount of data transfers to and from the computing units. MRAM could help reduce this by keeping data in one place instead of constantly transferring them between SRAM, DRAM, and main storage like flash memory in a traditional setup.

The recent spike in investment interest in potential AI plays has given MRAM, or its stock to be exact, a major boost due to its perceived need for AI. The chart above shows how the stock’s YTD gains have accelerated in recent months. The stock was up 14.2% for the year as recently as May 1, but the stock took off from then on to bring YTD gains to 87.4% after the stock closed at $10.42 on October 12.

MRAM took off in May, as did many other semis, which is no coincidence since May is the month the sector got a jolt after NVIDIA ( NVDA ) blew past earnings expectations due to AI. This sent many investors scurrying around to find other stocks that could potentially benefit from AI the way NVDA did. This is where MRAM comes in.

Why MRAM may be due for a correction, if not something more

However, the stock’s rise has slowed down. The stock reached a 52-week high of $10.50 on September 1, but there have been no new highs since then. In fact, the stock retreated afterward, although it did manage to make it back to the recent high. The stock hit $10.50 on October 10 and October 12 and it got within $0.05 or less of $10.50 on October 6 and October 11.

Notice how in the previous chart the stock seems to be stalling out in the $10-11 region. The stock got close to this region as early as June 30 with $9.99, but months have passed and no breakthrough has occurred. The price action points to resistance in the $10-11 region, based on the stock’s inability to get no higher than $10.50 despite repeated attempts. There seem to be a lot of sell orders at $10.50, which could be why the stock retreats whenever it hits $10.50.

It’s worth mentioning that the stock has been gradually going up for over a year, ever since the middle of 2022. This was in turn preceded by a fairly steep drop from late 2021 to mid-2022. An assumption could be made that the current rally in the stock, which started after the stock bottomed in July 2022 at $4.71, is a retracement of the downtrend that preceded the bottoming.

If this is the case, and assuming the downtrend started with the December 2021 high of $14.36 and ended with the July 2022 low of $4.71, then the closely watched 61.8% Fibonacci retracement can be found at $10.67. This is just $0.17 above $10.50 and within the aforementioned $10-11 region. This could explain why the stock is encountering resistance in the $10-11 region or why so many appear to have decided the stock is a sell in the $10-11 region.

In addition, the $10-11 price region may also be where fair value can be found. Fair value is a highly subjective term and there are different methods used, which can be up for discussion. Nevertheless, MRAM has made major improvements in terms of profitability in the years following its IPO in 2016. MRAM has improved the bottom line every year since then, going from a GAAP loss of $3.53 per share in FY2016 to a net profit of $0.29 per share in FY2022.

Granted, MRAM was in the red for most of the time, but it has earned a profit in each of the last 9 quarters. If we then assume MRAM continues to make rapid progress by growing EPS by an estimated 30-31% on average in the next five years and with TTM EPS of $0.34, then the fair value is around $10.20-10.54. This could explain why the stock has gotten no higher than $10.50 since any higher could put the stock far above fair value.

If the stock is trading at or slightly above fair value and the stock is close to an important resistance level, then it becomes easier to understand why the stock could have a hard time continuing the rally of recent months. This does not mean the stock cannot go higher, but the path of least resistance is likely to the downside.

Could earnings disappoint?

As mentioned earlier, MRAM has improved the bottom line every year since it went public, although from a very low base. In fact, the stock rally starting in May was assisted in no small degree by an earnings report that beat estimates and included guidance that was better than expected. MRAM, for instance, reported a GAAP profit of $0.04 per share in Q1 FY2023, much better than consensus expectations of a GAAP loss of $0.01 per share.

This continued in the most recent report. MRAM earned $0.18 in Q2 FY2023 as shown in the table below, which represents a sequential increase of $0.14. However, it’s worth mentioning EPS benefited from a $2M tax credit. Without it, EPS would have been more like $0.08, or $0.10 less than $0.18. Still, adjusted EBITDA rose QoQ and YoY, a sign of improved profitability. MRAM finished the quarter with cash and cash equivalents of $30.8M with no debt.

Source: MRAM

On the other hand, guidance was on the soft side with EPS of $0.01-0.06, down QoQ and YoY.

MRAM added some color to the outlook. Demand remains healthy overall, but MRAM has seen early signs of softness with some cancellations and pushouts. On the other hand, bookings are headed into H2 2024 and the backlog remains. From the Q2 FY2023 earnings call:

“From a backlog standpoint, the backlog continues to be very healthy, so I'll just kind of reiterate some of the things we've said in the past. We're seeing bookings now going into the second half of '24. There's new orders outside the lead times. And we're continuing to see increased bookings within the lead times. So from that perspective, things are looking good. We do continue to see some cancellations and pushouts, but that's limited and I would say probably very normal if you look at the historical trends. So things are looking promising from a backlog perspective.”

A transcript of the Q2 FY2023 earnings call can be found here .

Multiples may be an issue

MRAM trades at higher multiples than most semis, which may become an issue for some. For instance, EV/EBITDA is 19x on a forward basis and 32x on a trailing basis. In comparison, the median in the sector is 13.6x and 13.7x, respectively. The table below shows some of the multiples MRAM trades at.

Source: Seeking Alpha

Investor takeaways

Of all the different types of emerging non-volatile memory types looking to combine the best of DRAM and flash memory, MRAM is the one option arguably closest to being ready for prime time, mostly because MRAM had a head start on most due to it being developed back in the eighties. If someone needs persistent memory, which combines DRAM and flash, and cannot wait for something better, then MRAM is arguably one of the best, if not the best options out there. MRAM may have a number of weaknesses, but it is relatively mature and most importantly available. Simply being something that can be used immediately with relatively good performance puts MRAM at or near the top of the pack among emerging non-volatile memory types.

However, it’s worth mentioning that the use of MRAM has yet to really take off, whether in AI or elsewhere. Revenue more than doubled from $27M in FY2016 to $60M in FY2022, but an argument can be made that an increase of $33M in six years is not that spectacular. MRAM is still very much a niche technology, which is reflected in sales or the lack of it.

Keep in mind MRAM is not particularly new. MRAM was first developed in the 1980s and many companies like Micron ( MU ) have worked on commercializing MRAM through the years, only to abandon the technology for various reasons. STT-MRAM is newer, but it’s still just an improved version of MRAM. It may take a long time before MRAM truly takes off, if ever.

Furthermore, MRAM faces competition from alternative forms of non-volatile memory like resistive random-access memory or ReRAM/RRAM, ferroelectric RAM or FRAM, phase-change memory or PCM, and many others. The industry has yet to decide which, if any, is the best out of all the competitors. It is possible MRAM could lose out in a competition. Remember how 3D-Xpoint, another type of non-volatile memory seeking to combine the best of DRAM and flash memory, has been discontinued, even though MU and Intel ( INTC ) had high hopes for it and spent a lot of resources on it.

Nothing is set in stone, but one could argue that MRAM’s drawbacks like manufacturing complexity put it at a big disadvantage compared to other more recent types of non-volatile memory, especially as it makes MRAM the more costly option compared to less complex and thus likely cheaper options like say RRAM. MRAM may see limited uptake, but the combination of complexity, higher cost, and potentially lower reliability could be a burden too difficult to overcome.

MRAM is very much a speculative play. MRAM may take off, but it could also be superseded by a superior alternative, whether it is RRAM or whatever new technology appears. MRAM may be trading close to fair value and the stock could be close to resistance. The stock has had a long run, which means the rally is long in the tooth and MRAM is probably long overdue for a sizable correction, if not something more.

I am therefore neutral on MRAM. The decision as to whether MRAM is worth considering ultimately comes down to whether one believes MRAM can come out as one of the winners, or perhaps even the winner, in the market for non-volatile memory. If yes, then long MRAM is worth considering. Just be aware the race for non-volatile memory has yet to be decided. MRAM may be in the lead, but something can always come from behind to overtake it.

For further details see: