Exhibit 7.01
Board of Directors
GSI Technology, Inc.
1213 Elko Drive
Sunnyvale, CA 94089
Dear Board of Directors:
Roumell Asset Management, LLC owns nearly 2 million shares of GSI Technologies (GSIT), roughly 8% of the company's outstanding shares. Our persistent accumulation of the company's shares indicates one thing - we believe the company's underlying intrinsic value is materially greater than current price levels.
We take great comfort in being invested alongside visionary founders Lee-Lean Shu and Robert Yau, who together own 18% of the company's stock. Together with other senior management team members who own about 16%, aggregate executive ownership is roughly an impressive 34%. The company's engineers have done a first-rate job in combining MikaMonu's software (acquired in November 2015) with the company's hardware capabilities to create the Gemini platform. The engineering work has been underscored by the recent Israeli MAFAT first place win, the NASA win and, now, importantly, selection by AWS's OpenSearch initiative. GSIT's sales team is to be applauded for such high-profile achievements within a year of Gemini being introduced to the market. The board, on the other hand, has failed in its role, in our opinion. Accordingly, we voted against all board members at this year's annual meeting.
First, there is a complete absence of buying by directors, even among very long-tenured board members. Board members Jack Bradley, Haydn Hsieh and Ruey Lu have been on the board for 6, 13 and 21 years, respectively, but have not seen fit to make any open-market purchases over the past several years. In our opinion, the company's directors should be willing to vote with their pocketbooks by making open-market purchases to signal their belief in the company's enterprise. If they're unwilling, each should step aside and allow talented individuals, with a commitment to risk their own capital, to replace them. After Arthur Whipple's recent departure, board members Bradley, Cholawsky, Hsieh, Le and Lu beneficially own a paltry 140,000 shares in total, or roughly 1% of the company, overwhelmingly obtained through board compensation. Candidly, each should be embarrassed by their unwillingness to invest alongside the company's shareholders.
Second, we believe there is growing concern among GSIT investors that the board lacks capital markets sophistication. Is the board aware that the private equity capital markets are on fire and paying large sums of money to access AI chip opportunities? Recently, SambaNova Systems raised $676 million with a pre-money valuation of $5.1 billion. Groq raised $300 million for a pre-money valuation well over $1 billion. Cerebras Systems raised $270 million in November 2019 at a pre-money valuation of $2.4 billion. Habana Systems was purchased by Intel for $2 billion in December 2019, and Nervana Systems was purchased by Intel for $350 million in August 2016. Our understanding is that all of these companies were pre-revenue enterprises. In the meantime, GSIT trades at a market cap of roughly $130 million; after subtracting out the company's cash position of about $60 million, the company's enterprise value is a paltry $70 million, which is likely fully accounted for by its SRAM business that appears to be successfully pivoting to higher ASPs and margins as compared to its historical metrics. In addition, GSIT purchased its headquarters in 2010, which we estimate is worth about $15 million today. In other words, the company's AI Gemini platform, affirmed by the wins noted above, is inexplicably being given an implied valuation of zero within the backdrop of sophisticated investors who are hungry for semiconductor AI exposure.
In each of the examples noted above, the company raising capital provided its investors revenue projections and TAM information. Having been invested in GSIT for several years, we firmly believe such projections have been run inside the company. How could the company have purchased MikaMonu in 2015 and made the significant time and capital commitment to build the Gemini platform without such information? We call upon the company to share, to the best of its ability, such information with the markets. This is something the board should have demanded long ago.
An engaged board should do the following:
- Immediately begin identifying and engaging with potential strategic partners for NRE funding on Gemini. Benefits: reduces cash burn on Gemini R&D and could lead to potential customer introductions and other market opportunities.
- Instruct the CEO to evaluate opportunities for licensing the APU IP. Benefits: generates cash and provides potentially high-margin revenue.
- Conduct a valuation analysis on the current patent portfolio. Benefits: underscore and objectively understand the company's IP value for potential investors (private or public).
- Engage a financial adviser to perform a valuation analysis of the company both to attract strategic partners, furthering the first engagement initiative, and as a proactive measure should an unsolicited offer come over the transom. There needs to be a basis in place to compare a potential "bird-in-the-hand" offer to a "two-in-the-bush" potential future value.
In a May 5, 2021, Reuters published a piece titled, "U.S. chip start-ups, long shunned in favor of internet bets, stir excitement again." The author noted, "Silicon Valley venture capitalists, long focused on software and internet companies, are again pouring money into the semiconductor industry, lured by the promise of a new generation of artificial-intelligence chips that could challenge incumbents like Intel Corp. and Nvidia Corp." The author also stated that venture funding of U.S. chip companies reached its highest level in two decades last year at $1.8 billion and that venture capital had already invested $1.4 billion as of May 2021. There's good reason for Silicon Valley's interest in the AI chip space. Gartner's AI semiconductor revenue forecast for the sector estimates 2024 revenue of $61 billion, up from $34 billion in 2021, a 20% plus CAGR. Allied Market Research estimates the AI semiconductor chip market will reach $91 billion in 2025. Analytics Insight, referencing an IHS Markit report, indicates "potential expansion" in the AI semiconductor market to $128.9 billion by 2025.
An engaged board and management team would creatively think about ways to unlock value, while retaining long-term optionality. For instance, Gemini could be placed into a separate entity, allowing outside investors an opportunity to participate in, and possibly even contribute to, its upside potential. Currently, investors cannot effectively access Gemini's optionality, largely because of the company's meager float. A new vehicle, in which GSIT maintains a majority interest, would provide the potential to unlock substantial value under current market conditions, and at the same time retain a meaningful stake in Gemini's future for the company's shareholders.
As an example, earlier this year HCI Group, Inc. (HCI), put its TypTap business in a separate vehicle and sold a 12% interest for $100 million to Centerbridge Partners, thereby establishing an underlying value of $850 million for its TypTap asset. HCI stock traded at $58/share prior to the spin-out, and immediately appreciated to $75/share days later, and today, eight months later, trades over $100/share.
Engaged boards should think imaginatively and boldly on behalf of shareholders. GSIT's engineers and sales teams have done the heavy lifting; shareholders deserve a board willing to do its part in maximizing shareholder value. To be clear, we do not believe that the company should be sold outright today, as our view of intrinsic value is multiples of the current stock price. We call upon the board, and the executive leadership team, to immediately hire outside advisers to think through some of the options presented in this letter, and possibly others, and take advantage of flush capital markets desperately looking to invest in the AI semiconductor space.
Very truly yours,
/s/ James C. Roumell
President
Roumell Asset Management, LLC