Description

Everspin is an overhyped semiconductor company that I believe will miss sellside’s lofty revenue estimates for 2018. The stock stumbled out of the gate after its IPO in October 2016, but it rocketed from $10/share to $25/share in two months on the heels of Barron’s positive mention in May of this year.  The stock has retreated since its peak but is still 100% above the IPO price of $8/share.  I think there is 40-60% more downside as investors recognize management’s inability to meet 2018 estimates due to the limitation of its touted technologies. I peg downside valuation at $7-$10/share, or 2-3x 2017E revenue, vs 5x at current valuation.  Meanwhile the company continues to burn cash and will likely raise more cash in a secondary offering after its one-year IPO anniversary.

Company background

Everspin develops magnetic random access memory chips (MRAM), which uses the magnetism of electron spin to provide fast and enduring non-volatile memory. The benefit of MRAM is that it has the performance characteristics similar to standard Random Access Memory (though not as fast as DRAM) while having the persistence characteristics of Flash Memory (won’t lose data if power is removed). The primary use of MRAM today is in embedded applications such as automotive and industrial. The company was spun out of Freescale Semiconductor in 2008.  It went public in October 2016 at $8/share, below an expected $11-12 range.

Bull case

Today Everspin generates all its revenue ($31 million TTM) from its first generation MRAM products which have been shipping since 2008 to customers in industrial, automotive and transportation sectors.  However, Everspin believes the market for these products is limited and is betting its future on its third generation MRAM products for the enterprise flash storage market (SSD), which the company estimates as a $1.2 billion opportunity.

In a typical enterprise SSD system, the controller contains DRAM chips and capacitors/batteries to provide backup power so that data is not lost in DRAM in the event of power failure.  In theory, MRAM chips can replace the combination of DRAM and capacitors/batteries, thereby saving money and space and providing more reliability for the SSD customer. Everspin recently introduced its third generation MRAM products (Gen 3) to address the enterprise SSD market.

Another positive spin is that GlobalFoundries, which has licensed Everspin’s MRAM technologies, is likely to provide large stream of royalty revenue when it launches chips with embedded MRAM for its customers in late 2018.

Reality

Consensus revenue estimate for 2018 is $79.5 million, up 110% from $38 million expected in 2017. Management has already admitted that GlobalFoundries will not generate royalty revenue for MRAM until 2019 at the earliest, assuming GlobalFoundries is able to develop and market embedded MRAM solutions. I do not believe Everspin can deliver anything close to consensus 2018 estimates because its Gen 3 MRAM products will not garner significant revenue for two reasons:

1. Everspin’s Gen 3 production timeline does not match sellside revenue forecasts.  Everspin currently has two Gen 3 products: 256Mb and 1Gb. The former will be available for delivery in Q4 2017 while the latter is in the sampling stage with production expected by end of Q1 2018.  However, on its recent Q2 call, management implied that customers are skipping 256Mb in favor of 1Gb. Moreover, management indicated that it would take 9-12 months from design-in to production ramp, which implies revenue contribution from 1Gb will likely not materialize until 2H18 at the earliest.
2. Industry participants do not believe MRAM will gain wide adoption in enterprise SSD. The most common reasons cited are performance, density, cost and ecosystem support. I recently attended Flash Memory Summit in Santa Clara and gathered feedback from attendees from various enterprise storage vendors. They were unanimous in their skepticism of MRAM:

·        Memory solution provider #1 (also Everspin reseller): “The problems with MRAM are density and cost. Don’t see demand. Need at least 1Gb. Also too expensive. Could see demand in a year or so but would take a while.”

·        Memory solution provider #2: “I don’t see demand to replace DRAM. Batteries and capacitors are not an issue. MRAM can’t come close in density and performance.”

·        All-flash storage vendor: “We are not considering MRAM because we need DRAM’s speed. Space for battery backup is not an issue, nor is the cost of DRAM and capacitors when looking at the overall cost of the storage system.”

·        Enterprise storage vendor #1: “We have a prototype of MRAM-based system, but we are not seeing demand. The issues are cost, performance and density.”

·        Enterprise storage vendor #2: “It will be tough to adopt MRAM because most server BIOS don’t support it.”

·        Enterprise system provider: “I can see MRAM for niche use but not for enterprise SSD due to performance and density problems. I don’t see how Everspin can scale the technology. Capacitor space is not an issue.”

Risks

MRAM announces Tier 1 Storage Customer: Everspin recently replaced its CEO with the former CTO of SanDisk. I don’t doubt that the new CEO can use his rolodex to get meetings with his network of industry contacts.  He may even get a design win or two, causing the stock to pop in the short-term.  However, given the industry feedback I have received above, I do not believe any of these announcements will result in significant revenue opportunity.

Catalyst

Continued cash burn: Everspin has been burning cash over the past 3.5 years. Management expects to burn $4 million per quarter until its Gen 3 products are ramped.

Secondary offering: Due to its cash burn and weak IPO offering, Everspin has only $9 million in net cash on its balance sheet. The company will need to raise additional cash soon.  Management has hinted that it may do a secondary offering in October after its first year IPO anniversary.

Miss 2018 estimate: I expect Everspin to miss 2018 consensus estimates due to lack of market acceptance for its Gen 3 MRAM products.