GSIT - GSI Technology: Gemini-II On Track And Cash Burn Slated To Fall

2023-04-11 08:00:47 ET

Summary

• GSIT says they'll be doing around $7 million in lower costs to reduce cash burn, giving more than a two year horizon easily for cash burn.
• The Gemini-II is on track and GSIT expects orders to possibly start after summer.
• They are focusing their R&D on the SAR applications, which means that GSIT is going to get its start in primarily military markets beginning with Elta.
• APU should beat out the GPU in the Elta benchmarking test, in which case real sales could then start to Elta as a first customer. This will be the first commercial proof-of-concept moment.
• Assuming that all the tests and competitions they've won in other applications with the APU were legitimate, there really is nothing negative on this story except for the cash burn and dilution risk.

AI is definitely the flavour of the day with ChatGPT. If you are scouring markets for an advantaged AI play, we maintain that GSI Technology ( GSIT ) remains one of the best asymmetric opportunities on markets for investors considering the AI factor for their portfolios. Indeed, recent speculation on AI has helped the stock recover, albeit to still very depressed levels. Until the last couple of years, their legacy businesses provided a business option for them to explore the APU, the technology for which was acquired from an Israeli CS Ph.D's company. They won competitions with it, beating out competition with a revolutionary architecture, and this year they are finally getting their MVP into high-profile commercial hands as a proof of concept. Cash burn is falling on cost cutting reducing the dilution risk concern, and their approach to a commercial stage with first tests of their commercial grade chip starting in 2024 could become a major catalyst. Strong buy, but speculative.

Q3 Call

Let's walk through the key elements of the Q3 call, starting with the financial considerations. When asked about cost-cutting, the CFO said:

Well, we were looking at somewhere around over $13 million a year and the cost cutting will save us about $7 million a year. So we'll be better off than we were a year ago I believe.

Douglas Schirle

Last year's cash burn was around $13 million , this year will line up similarly, and while some cost elements will grow as they continue to ramp-up work on the new Gemini-II, cash burn should not be expected to grow. They fired about 15 people bringing the headcount down from 180 to 165. Revenues were still decent this quarter despite no further Rad-Hard prototype deliveries, where Rad-Hard was expected to come in an potentially double revenues from the legacy SRAM business . We should expect decent SRAM performance going forward, and with the eventual Rad-Hard-related satellite launch a jump in revenues at some point as well.

Given the expected rate of cash burn, from December there should be over two years before any equity raise would be needed, consistent with previous management comments. We should also note the total lack of debt on the balance sheet. While credit conditions are worsening, there is debt capacity despite overall cash burn due to the ability of the SRAM business to produce positive cash flows, which it has been doing till now to help self-finance the Gemini R&D.

There were more things to consider in the call. An investor asked a question about miniaturisation opportunities, and management addressed this question by pointing out that without the Gemini-I clunky FPGA, there is substantial scope for the Gemini-II to be applied to smaller form factors. In general, massive cost savings can be achieved from an industrial point of view with the development of the next Gemini-II to take the revolutionary architecture but apply it better. 30x cost performance improvements and 8x memory density is what the Gemini-II brings to the table, essential for the next phase of commercialising the product, which could start with Elta.

The last quarter signaled the most relevant development in years which is that after years of winning competitions with the APU, a major commercial customer from the Israeli aerospace industry is wanting to use the Gemini-I and test is for SAR applications, which is the creating of high resolution images from low-res radar data using pretty intensive algorithms that will benefit from in-place operations possible with the APU. This SAR technology has a decently sized market across military players, and all the effort by GSIT's shrunken headcount is going into developing the software needed to equip Elta to get the APU to work for them for that application, as opposed to trying to expand applications for the APU at this point which they were likely doing in previous quarters with larger headcounts, since headcount cuts have been coming from R&D which is down 10% YoY in expenses. This focus on making the MVP work with software for their first commercial catch makes a lot of sense. Apparently the only thing standing in GSIT's way is a GPU that Elta will be comparing the Gemini-I against. With superb performance in competitions beating out GPUs, whose architecture is fundamentally less suited, the Gemini-I could win and Elta could be the first customer ordering APUs in late 2023 and hopefully creating sales from the Gemini-II in 2024 - we may get some minor sales of the Gemini-I already in 2023.

Bottom Line

The Elta news is key to follow as we wait for the Q4. If the APU cannot beat out a GPU, this investment is rather dead. It takes a while to get things to work, and being a small player like GSIT means that it's hard to get these commercial arrangements into place. Moreover, if the APU loses to a GPU that would be concerning for the effectiveness of the APU in general. Things can really go wrong for GSIT, and of course this project has burned a lot of cash. With dilution being concerning for investors since the GSIT price has retreated so much, the risks are quite high with this stock, and investors should reflect on how to size a position in a stock like this correctly.

We also want another confirmation that the Gemini-II is indeed on track, since this is critical for them to be able to scale.

However, assuming that the APU performs, dilution risk isn't so high thanks to cost-cutting measures, and the APU could, in principle, be central to lots of major machine learning applications. For investors looking to get an AI factor into their portfolio, GSIT could be the company that has the technology to produce among the most important technologies for the AI revolution. We remain long and label it a strong buy, albeit a speculative one.

For further details see: