I was intrigued with William’s find of ONTO. I spent the first two years of my career (1987-1989) working for Shinko Electric, a semiconductor packaging manufacturer that is still in operation today. ONTO makes packaging inspection equipment for semiconductors that includes combinations of SOC (systems on a chip that consist of GPU or CPU or other devices) and HBM (high bandwidth memory), the RDL Interposer (aka the wiring that connects the SOC and HBM), the C4 (solder), and the PCB (printed circuit board). The figure shows how the packaging connects everything. This interconnected system is critical to reduce power requirements for AI datacenter, wearable devices, and smartphones. In addition, ONTO also recently acquired a company used in metrology for electronic devices used for electric vehicles and high power devices.
ONTO’s business falls into two basic categories, upstream equipment, which is about 84% of the revenue, and software subscriptions that go with the equipment, which is about 16% of the revenue.
ONTO seems to be at an inflection point where orders are ramping up. By ONTO’s own account, orders are LUMPY, but as far as they can see revenue is ramping up fast and will remain strong many quarters into the future. ONTO’s backlog has reached a record $1.1B. Revenue growth should be accelerating beyond the 35.3% seen this quarter, margins are expected to be slightly above current levels but won’t see the 350 basis point jump like they achieved this quarter.
What is ONTO’s secret sauce? It provides industry leading metrology and etching devices for microchip packaging. During the assembly of the semiconductor package, microchips (aka dies) can shift ever so slightly and the RDL interposer or the PCB board can warp ever so slightly, and inspection equipment that ONTO makes sure that the assembly is functioning properly in its final state while maintaining a clean, low moisture environment for the chips.
To be interested in ONTO, you need to believe that the AI chip boom will continue. I subscribe to this theory because NVIDIA claims that data centers will be paying for themselves going forward and memory makers are building new foundries based on long term commitments for their chips. If data centers do pay for themselves, I think this might avoid the capital concerns that we have seen to date. What do I plan to do? I plan to compare them to Figure Technologies and Bloom Energy If I do invest in them, I’ll keep them on a hair trigger. Any slight chink in the armor of my thesis, and I will be out of ONTO faster than it takes two jack rabbits to have a date. Equipment makers are natoriously boom or bust, though the software component of their business may soften any blows.
