2026.09.22 | Substack | AMBA | Gaetano: AMBA Physical AI Position

AMBA is one of my largest Physical AI positions, and I am really holding this one for what I think the business can look like in 2028.

Physical AI Model Portfolio

Gaetano

Jun 1

Ambarella provides low-power AI compute for robots, drones, cameras, vehicles and other physical systems. As those machines get smarter, I expect more of them get deployed, and the amount of compute inside each one increases.

We’ve gotten a few nice updates as of late since I first wrote about them. Their robotics design-win base has continued to grow beyond the 15+ wins and $100M+ lifetime revenue opportunity it disclosed earlier this year. It has also opened a much larger edge infrastructure opportunity, where its AI compute can be added alongside processors already sitting inside industrial systems rather than requiring the entire machine to be redesigned around an Ambarella chip.

The route to market is expanding too with new relationships that could eventually open AMBA to a much larger group of industrial and enterprise customers, while its first 2nm semi-custom AI program is expected to begin generating production revenue in FY28.

That gives me more ways to get to the 2028 outcome than I had when I originally bought the stock.

After the paywall, I want to focus on what I think AMBA could actually earn in 2028, what multiple I think that deserves, what I think the stock is worth today, and what it could be worth if the thesis plays out.

I think AMBA is worth roughly $80 per share today. My current model has calendar 2027 revenue around $550M, and I am comfortable using roughly 6x EV/Sales on that number. That multiple sits between some of the more conservative sell-side work around 5x and Stifel’s current 7.6x framework. After adding back AMBA’s net cash and accounting for the share count, I get to right around $80. With the stock around $66 and my average price just over $70, I like where I am positioned.

The bigger reason I own it is what the business could look like as we move through 2028. If robotics continues converting into production, edge infrastructure starts scaling, semi-custom contributes, and the new sales channels develop the way I expect, I think AMBA can move toward roughly $750M to $800M of annual revenue. If the business is still growing above 20% at that point, I think roughly 6–7x sales is reasonable. That would support something around $110 to $130 per share, with more upside if growth ends up closer to 30%.

My downside case is around $50. That would mean the newer businesses take longer to develop, revenue stays much closer to the current Street trajectory, and the market values AMBA closer to 4x sales rather than giving it a premium AI-growth multiple.

Financial Disclaimer

This article is for educational and informational purposes only and reflects my personal research, estimates, assumptions, and opinions at the time of publication. Nothing in this article should be considered individualized investment, financial, tax, or legal advice, or a recommendation to buy, sell, hold, or short any security.

I currently own shares of Ambarella (AMBA) and may buy, sell, add to, reduce, or exit my position at any time without prior notice.

Any revenue estimates, valuation multiples, price targets, fair-value ranges, or other forward-looking assumptions discussed are inherently uncertain and may prove incorrect. Actual results can differ materially from my expectations. Investing involves risk, including the possible loss of principal.

Please conduct your own research and make investment decisions based on your own financial situation, risk tolerance, time horizon, and objectives.