MCHP - Superb Execution And Growing FPGA Usage Driving Lattice Semiconductor To New Heights

Summary

• Lattice delivered a little upside to sell-side numbers in Q4, but also delivered sequential growth in a quarter where most chip companies saw contraction; guidance calls for no contraction.
• I believe Lattice is benefiting from MCU shortages, growth in new/expanding use-cases (like machine vision), and efforts to make its own FPGAs easier to program.
• Expanding into mid-range FPGAs this year will double the addressable market at attractive margins.
• Lattice already trades at or beyond comparable M&A take-out prices; while low-power FPGAs may be a game-changing product, it's hard to make any valuation call here.

The Lattice Semiconductor ( LSCC ) rocket ship is soaring ever higher, climbing almost 70% since my last update (in November!), as the company continues to execute exceptionally well, while also leveraging strong underlying growth in FPGA demand. With microcontrollers (or MCUs) still capacity constrained, growing use cases for edge processing (particularly in situations where low power consumption is invaluable), and Lattice continuing to move the ball forward in terms of product and software design, almost everything is going the company’s way.

Valuation is pretty much a pointless conversation now. I can’t really find a context in which Lattice is cheap, other than to note that the company continues to exceed lofty expectations, while also expanding its addressable market.

A “Bad” Quarter That’s Still Very Good

One of the hallmarks of great growth stories is that they’re practically immune to macro pressures. Not only did Lattice not see the sequential revenue decline that many excellent semiconductor companies saw, I don’t think they’re going to see any sequential shrinkage through this phase of the cycle. Low single-digit growth may be as bad as it gets, as the company continues to benefit from strong market dynamics and its own product design and marketing/commercialization efforts.

Revenue rose 24% year over year and 2% quarter over quarter, coming in just ahead of sell-side expectations. Gross margin improved almost five points from the year-ago quarter and 50bp sequentially, reaching 70% (for those who don’t routinely follow semiconductor companies, this is exceptionally high). Operating income rose 52% yoy and 3% qoq, with operating margin rising 730bp yoy and 50bp qoq to 40.2%.

Inventory days did jump in the quarter (up around 65 days yoy and 28 days qoq to 187 days), but that is a byproduct of the need to manage (build) inventories ahead of a wider launch of its new mid-range Avant products and to support growing use of its products in new applications.

Auto and Industrial sales were exceptionally strong at +51% yoy and +4% qoq, outperforming peers in the analog space. Auto demand continues to be driven by infotainment, ADAS, and sensor aggregation, while the company sees strong industrial demand from automation, machine vision, and robotics. Communications and Computing grew 15% yoy and shrank less than 1% qoq, with some near-term weakness in servers and PCs. Consumer declined 17% yoy but rose 2% qoq; this is a surprising solid result given the weakness in consumer IoT/smart home flagged by companies like NXP ( NXPI ) and Silicon Labs ( SLAB ), but content wins could easily explain the outperformance.

So, What’s Going On?

I believe there are multiple drivers of Lattice’s success, most of which support a strong ongoing multiyear growth story.

Starting with what may be the shortest-lived driver, I believe Lattice could well be picking up business at the expense of capacity-constrained MCU manufacturers like Microchip ( MCHP ) and NXP. Auto MCU lead-times remain highly elevated and NXP has said they’re basically sold out of MCUs for auto applications through the end of 2023. While I have no doubt that Lattice is winning auto sockets meant for its low-power FPGAs, it’s at least plausible to me that customers are opting to use these chips as a temporary stop-gap for MCU shortages.

While that is a threat down the road (as MCU capacity ramps up), multiple MCU manufacturers have suggested that capacity constraints could last for years, as companies are reluctant to build capacity of their own and foundries aren’t prioritizing the older nodes where most of these MCUs are made.

A stronger driver is simply the capabilities of FPGAs relative to other options. In edge-based imaging (like security cameras, machine vision, presence detection, et al), FPGAs can process images much more quickly as, like GPUs, they can do parallel processing. That speed is critical for applications like scanning a fast-moving assembly line or trying to capture and process faces in a crowd or license plates on a highway, and conventional processors are too slow. On the other hand, GPUs are power-hungry, making FPGAs a more efficient option when the processing capabilities are similar.

I also believe that an underrated aspect of the Lattice story is the effort spent by the company to develop software tools. FPGAs have always been appreciated for their flexibility and reprogrammability (the FP means “field programmable”), but programming them can be a very complex task. With a suite of tools designed in conjunction with their chips (and with an eye toward specific tasks), I believe Lattice has significantly reduced the burden of using their chips in place of MCUs or other options.

Now with the company about to launch Avant, its new mid-range product (built on 16nm FinFET w/ TSMC ( TSM )), the company is basically doubling their addressable market and moving more toward Microchip’s wheelhouse in mid-range FPGAs. I’d also note, though, that the company has been winning some slots in areas like enterprise networking that you’d typically think would be the territory of Intel ’s ( INTC ) or AMD ’s ( AMD ) high-end FPGAs. I’m not suggesting that Lattice is moving into high-end FPGAs, but rather that use cases are expanding to a point where they’re seeing wins in market categories where low-range FPGAs haven’t historically seen much traction.

The Outlook

Business has continued to develop pretty much as I’d hoped, so I’m not really changing much in my model. I’m looking for 16% revenue growth next year, putting me near the top of the sell-side estimate range, and my 20% growth estimate for FY’24 has me about 5% above the highest published sell-side estimate today. Longer term, I’m looking for mid-teens revenue as use-cases for low-range and mid-range FPGAs continue to emerge and expand across auto, industrial, communications, and computing end-markets and as Lattice continues to gain share.

I’ve modestly tweaked my margin assumptions higher, but not by what I’d consider a material amount, and I’ve adjusted some of my working capital assumptions (which impacts near-term FCF). As far as core assumptions go, nothing has changed – I’m still expecting low-to-mid-30%’s adjusted FCF margins over time, driving high teens annualized FCF growth over the next decade-plus.

Valuation is … problematic.

Lattice already trades above the multiples where Altera and Xilinx were taken out (and I expect Lattice to reach the size Altera was when it was bought out during mid-FY’25). Lattice also trades beyond the 42.5x forward EPS multiple where Marvell ( MRVL ) acquired Inphi, a fast-growing interconnect specialist that dominates the PAM-4 market. Relative to my own estimates, Lattice trades just under that 42.5x forward P/E multiple (42.2x), and around 17x my forward revenue multiple.

Maybe there are comps here that I’m forgetting about or not aware of, but I can’t think of too many scenarios of valuations above and beyond this. I fully grant that Lattice is a special case and that the market opportunity for low-power FPGAs in a world of growing automation and electrification is huge, but I just can’t tie this back to any sort of reasonable multiple or valuation framework.

The Bottom Line

Sometimes, every now and then, you see a company that breaks the rules and you just have to ride with it. I can’t pretend that I don’t wish I’d gone ahead and bought shares back in November, and I may think the same thing about today’s price in another six months. I love the underlying fundamental story at Lattice and I’m not prepared to call a top; valuation seems very stretched, but I know the market will pay dearly for this kind of growth story. Personally I’ll wait for another pullback, but I may not get the opportunity I want, and that’s just how it goes when you try to invest with some semblance of value/price discipline.

For further details see: