Description

Introduction

I believe Super Micro Computer (SMCI) currently offers an attractive risk/reward special situation opportunity. SMCI has grown rapidly over the past few years to become a leading non-branded server supplier to the general enterprise, cloud, emerging storage and high performance computing spaces, growing 10x over the last 12 years from around $300M in annual revenue in FY June 2006 to over $3B in revenue expected in FY 2018. An Intel server processor upgrade cycle is driving an acceleration in growth that should last the next few quarters, with margin tailwinds from expected reductions in memory prices after that. Despite strong current revenue growth and the potential for future margin expansion, the stock trades near multi-year lows due to an accounting inquiry which I expect to be cleared up in the next couple of months. If the company can successfully bring its financials current, and analysts update their estimates to reflect the current business strength, the stock could double or better in the next year. Against this potential reward, the company currently has approximately $15.50 of tangible book value which should cushion losses in the event that the accounting inquiry takes longer or requires a significant restatement.

Business Overview

SMCI describes itself as “a global leader in higher performance, high efficiency server technology and innovation”. Products range from complete servers, storage modules, blade servers and workstations and other computing form factors. SMCI describes its products as offering “a high degree of flexibility and customization by providing what we believe to be the industry’s broadest array of server configurations from which they can choose the optimal solution which fits their computing needs”. The company offers over 5,000 SKUs, and according to the FY 2016 10-K sold to over 800 customers in 100 countries. The company uses third party suppliers and contract manufacturers, but does assembly, test and quality control in its own manufacturing facilities in San Jose, the Netherlands and Taiwan.

Obviously the basic server market is a slow-growing commodity market dominated by large suppliers like Dell and HPE with ODMs like Quanta increasingly getting into the game by supplying hyper-scale customers like Facebook and Google with servers of their own design. SMCI has carved out a niche between these two by providing customized, high performance server and storage designs to enterprises and data centers who want the latest technology customized to their own needs at a lower cost than the solutions sold by HPE and Dell. SMCI also serves a number of sub-niches in the server market, especially by providing the hardware backbone to newer data center product companies such as Nimble Storage and Nutanix. It has also gotten into other areas like small form factor edge servers for the IoT market (about 10% of sales), high performance computing and cloud computing- they supply IBM’s Softlayer division, which during past fiscal years has been a 10% customer.

The term “moat” in my opinion is one of the most over-used terms in security research, and I will admit it is tough to define SMCI’s “moat”. I believe they win business because they are aggressive in designing custom form factors and high performance systems. The company spent around $130M in R&D in FY 2017. They are particularly adept at quickly incorporating Intel’s latest processor designs into products, an important point that will be further discussed. CEO Charles Liang founded the company in 1993 and still owns approximately 18% of the company. To go from nothing to $3B in sales while staying consistently profitable and needing little outside financing he must be doing something right!

Current Business Trends

As mentioned above, SMCI’s engineering nimbleness enables it to quickly incorporate Intel’s latest server processor designs into a wide range of server products optimized for its customers’ specific needs. Intel releases a new server processor family approximately every 18 months in a “tick-tock” cycle, with the “ticks” representing significant fundamental architectural changes and the “tocks” representing incremental improvements. While SMCI has outgrown the server market over time, in the past the business has shown significant revenue growth outperformance during the “tick” cycles as it gains share during new architecture rollouts. Intel is in just such a “tick” processor cycle with Skylake released at the end of Q2, although according to my discussions with SMCI Skylake-based systems were only a mid-single digits portion of the September quarter revenue and a double digit percentage of December revenue. Therefore, I believe the next three quarters are the sweet spot of the Intel processor cycle for SMCI.

What’s impressive about SMCI is that prior to the Intel processor cycle really kicking off growth had already been accelerating. While final results have not been reported due to the accounting inquiry, the company originally reported 37% Y/Y growth in the June 2017 quarter, and a preliminary growth rates of 28% in the September 2017 quarter and 29% in the December 2017 quarter at the low end of the revenue ranges given. For the March 2018 quarter, the company guided to a wide range of $700M to $780M in revenue (11-24% Y/Y growth) but keep in mind that that they originally guided to $725M to $780M in revenue for the December 2017 quarter, only to report a revised range of $840M to $850M on January 30th, 2018. While SMCI’s growth comps are a bit tougher than usual, I believe they can achieve at least 20% growth in the next three quarters on the back of the Skylake cycle which will put them on track to generate over $3B in revenue in the June 2018 FY and $3.5B in the June 2019 FY. During the last major Intel “tick” server processor cycle in FY 2015, SMCI grew 36% on the back of growing 26% in the year before and still grew 11% in the following year.

While SMCI’s recent revenue growth has been impressive, it is not all roses in their financials. The company has suffered from the significant increases in both DRAM and flash memory prices during calendar 2017. DRAM and flash pricing grew substantially during 2017 as demand growth outstripped supply. To anyone who has observed such cycles and the semi capex market in the past, it is beyond obvious that the good times in memory are not going to last (witness MU’s 4.5x current year P/E). Samsung in particular is spending record amounts of capex to bring new memory supply online, and headlines as I write this discuss the first DDR-4 DRAM memory production by a Chinese-owned vendor. History teaches us that in memory cycles the price eventually declines to the marginal cost of production until memory vendors start losing money, stop spending on capacity expansion and wait for demand growth to catch up. Indeed, during 2017 flash prices have already turned down although DRAM prices remain stubbornly high and experts expect DRAM to remain in undersupply until the end of 2018. Memory prices affect SMCI in that they are purchasers of server DRAM and flash-based solid state drives, and DRAM is a larger part of COGS than flash. Indeed, GMs have declined from 16% in FY 2015 to 15% in FY 2016, 14% in FY 2017 and the low 13% range in the December 2017 quarter.

The lower GMs have translated to lower pro forma operating margins, which peaked at around 8% in FY 2015 and have been in the 4-5% range as of late. I believe that the easing of NAND this year followed by the easing of DRAM as we get to the end of calendar 2018 will provide a tailwind to margins even as revenue growth slows after the Skylake cycle. Indeed, my financial projections have FY 2019 revenue growing at only 12% to $3.5B, and gross and operating margins of approximately 14.5% and 6.5% respectively. This should translate to pro forma EPS of $3.23 per share, and I think after two fiscal years of double digit growth and margin expansion a 15x multiple or stock price of $48 or so is not out of the question. While this may seem like a crazy amount of upside, I would note that the stock peaked at over $40.00 in the last major Intel processor cycle when the company was only about 55% of the revenue size I expect in FY 2019, and the company has traded at peak NTM P/E multiples of over 17 in the past. It should also be noted that due to the accounting inquiry many analysts have stopped updating their estimates so that many investors are getting an outdated look at SMCI’s fundamentals. Indeed, according to Thomson Reuters analyst consensus for the already reported December 2017 quarter is $797M vs. an updated range of $840M-$850M reported, and for revenue of $725M in the March 2018 quarter vs. guidance of $700M to $780M that I believe is significantly conservative.

Accounting Inquiry

Ah, but the accounting inquiry. SMCI held its fiscal 2017 Q4 earnings call as normal on August 3rd and presented a complete income statement, balance sheet and statement of cash flows. However, on August 29th SMCI filed an SEC form 12b-25 that indicated it would not be able to file its FY 2017 10-K on time without unreasonable effort or expense. On September 14th the company announced it received a letter of non-compliance from NASDAQ related it its inability to file its 10-K. When it reported preliminary financial information for fiscal Q1 on October 26th it revealed the reason that it was unable to file its FY 2017 10-K: 

“In connection with the in-process audit of the Company’s financial results for the year ended June 30, 2017, a sales transaction was subject to additional inquiry and review. The transaction in question was originally recorded as revenue during the quarter ended December 31, 2016. However, prior to review by the Company’s independent auditors and prior to the Company’s public announcement of its results for the quarter, the recognition of revenue was reversed and the revenue was subsequently recognized in the quarter ended March 31, 2017. When the audit committee was made aware of this transaction, the Audit Committee of the Company’s Board of Directors initiated an independent investigation to determine whether there were any similar transactions and if so, whether such transactions were properly accounted for. Due to the volume of data to be reviewed to complete the investigation, the Company was unable to file its Form 10-K on a timely basis.”

On November 10th the company submitted a compliance plan to the NASDAQ market. On January 16th SMCI filed an 8-K that included an extension letter from Bank of America that extended the date by which the delivery of the company’s financial statements could be made until March 13, 2018. I believe they would not have made this agreement unless they believed that they could bring their financials current by then. However, there is nothing stopping them from delaying further and getting another extension from Bank of America. On January 30th the company reported its preliminary Fiscal Q2 results and disclosed that the audit committee had completed its investigation. SMCI further stated they needed more time to analyze the impact of the results of their investigation and did not provide a date as to when the 10-K would be filed and whether historical financial statements would be restated.

Despite the better-than-expected preliminary revenue results, SMCI has traded off over 20% since the latest announcement. I believe this was partially due to the company declining to re-iterate that financials would be current by March 13th. Additionally, the company announced that longtime CFO Howard Hideshima was resigning and appointed employee Kevin Bauer as the new CFO. Their SVP of International Sales Wally Liaw (also a board member/co-founder) and SVP of Worldwide Sales also resigned. I believe that the resignation of Wally Liaw in particular also spooked investors as he has been with the company since its founding in 1993.

How Worried Should We Be?

To me, a current investment in SMCI basically turns on one question: is the accounting inquiry likely to be significant or not? As discussed above, I think based on how their business is trending and likely to do over the next 18 months over 100% upside is possible. At the same time, the company’s preliminary balance sheet as of 6/30/2018 showed $15.10 of tangible book value per share. As the value investors on this board have no doubt noted, there are not a lot of growing and consistently profitable companies trading at TBV. So I would submit that $15.00 is a good measure of the long-term downside, although that figure is certainly subject to some revision and the stock could certainly trade below that during an extended statement process.

With 150% potential upside and 20%ish long-term downside, the market seems to be saying that the accounting problems here are significant and there may be serious issues under the hood. Obviously I do not agree and this is the crux of why I believe SMCI is currently a great risk/reward opportunity. Below are the primary reasons why I think this accounting inquiry will not be overly significant:

• SMCI has been through this before. If you look back into SMCI’s filing history, it also filed an NT-10K in August of 2015. This time, the company needed more time to investigate irregularities regarding certain marketing expenses. While in this case the 10-K was originally filed shortly thereafter, the company was also later forced to amend its 2015 10-K and its fiscal Q1 2106 preliminary results between when originally released October 22nd 2015 and November 16th 2015 due to not properly deferring the value of warranties. The total amount amended for the FY 2013-2015 years was $9.3M of additional revenue deferred. This was no big deal, and I think is indicative of a company that has grown faster than its accounting staff’s capability had. Howard Hideshima was CFO back then as well, and I think the latest shortcoming in financial reporting was the final straw in his ouster.
• The transaction in question was deferred, not recognized. If you carefully read their original statement about what launched the latest accounting inquiry, the company stated it was around a single transaction where the product was shipped before calendar 2016 year end but the customer did not want delivery until after year end. Before reporting the quarter’s results, the company had deferred the revenue into the March 2017 quarter, but the way the transaction was documented led to the review of similar transactions. To me this points to more of a controls and procedures problem that a pervasive problem of improperly aggressive revenue recognition.
• Company does not fit pattern of “make the numbers” culture. For SMCI to be worth a lot less than the last stated TBV, I think it would have to emerge that the company had engaged in significant accounting fraud. Having seen a number of these cases in my career, I don’t think SMCI fits the pattern of a culture of fraud. Like many tech companies, they endeavor to give conservative guidance and beat it. However, SMCI has been forced to pre-announce results negatively more than once. They show significantly fluctuating growth rates and margins, not the metronomic “beat the quarter” steady results of accounting manipulators. The CEO is a quiet engineering type, not a showboating prima donna CEO who makes grandiose claims.
• Relatively simple business and accounting. Along with a “make the numbers” culture, accounting frauds I have seen in the past typically have complex businesses and accounting with significant M&A, non-cash transactions, multiple element revenue arrangements, cash burn, etc. SMCI has very little of this and is a relatively simple book-and-ship hardware business. For this reason I think there is relatively little scope for monkey business and any restatement is going to be minor around when revenue is recognized for a few transactions.
• Continued communication. During the period of time since SMCI’s 10-K was delayed, the company has continued to update investors on its results and even met with investors at the Needham conference in January. While it was disappointing that the company did not formally commit to the March 13th deadline in the Bank of America extension 8-K during its last conference call, they also announced that the audit committee investigation was complete. If there was really bad news to drop, I believe it would have happened then, and now the company is just working on the details. The continued communication by the company instead of just going dark has led me to believe that the issues are relatively minor.

To be sure, SMCI did report negative cash flow despite accounting profits in FY 2017 in its preliminary numbers, primarily due to a buildup of working capital. At 92 days the DSIs were at the high end of historical ranges, but understandable given the need to pre-purchase memory in a tight environment. DSOs at 61 days were slightly higher than the historical range of mid 40s to 50s but not the kind of level you’d expect to see at a company with significant revenue recognition problems. SMCI has reported years of accounting profit but negative FCF due to working capital buildup in the past, and the situation reversed in subsequent years. I think this working capital buildup plus the resignation of three officers is the reason the market seems to be treating SMCI’s accounting problems as serious, but I still believe that significant fraud is unlikely. I also think the officers’ resignation was due to failure to properly implement and follow proper accounting documentation procedures, not systematically over-recognizing revenue, but only time will tell.

Conclusion

For those who follow the tech industry, I think SMCI’s accounting inquiry is likely to look more like MRVL 2 years ago than any of the long drawn out sagas like Comscore, Global Eagle Entertainment, Avid or Comverse. For this reason, I think SMCI currently provides an excellent risk reward with 150% upside and limited chance of being worth less than $15.00. If the company can get current with its financials by March 13th or even in the next few months afterward, investor focus should shift to SMCI’s accelerating growth and the prospect for improved operating margins as memory supply/demand eases. If there are more significant accounting problems and there is an extended restatement cycle, an investor should be protected in the long term by SMCI’s $15.00+ in tangible book value.

Disclosure: The fund that I work for owns shares of SMCI. We may buy or sell shares of SMCI at any time without notice.

Catalyst

• Bringing financials current, possibly by mid-March
• Intel Skylake server processor cycle
• Declining DRAM prices in 2019