单片电源系统

Monolithic Power Systems' second-quarter results were strong, and third-quarter guidance was even better. Sales rose 48% year over year to $981 million, and guidance calls for 56% year-over-year growth in September. Management raised its enterprise data growth guidance to 130% in 2026, up from 85%.

Why it matters: MPS is capitalizing on immense demand for artificial intelligence and broad strength across its markets. We see further upside to data center revenue, which was initially guided to 50% in 2026. Strong results reflect best-in-class power chip design to us, and MPS moving up the value chain across markets. MPS is succeeding throughout the data center, with power solutions for GPUs, CPUs, memory, and optics. We like its strong market share and see robust investment in the medium term driving further growth. We also like long-term growth initiatives into rack-scale power and other new sockets. The non-data-center markets, such as automotive and industrial, continue to see nice growth as well, reflecting MPS' diversified opportunity. We continue to see the firm firing on all cylinders, even as AI and data center sales show the flashiest growth.

The bottom line: We raise our fair value estimate for wide-moat MPS to $1,650 from $1,500 to reflect higher growth estimates for 2026 and 2027. Shares rose 10% in the aftermarket but remain attractive to us following a tech selloff over the past month. We now model 50% sales growth in 2026 and 30% growth in 2027. This is data center-led but supported by all of MPS' end markets rising in our model. MPS remains an excellent play for long-term investors, giving huge AI growth exposure alongside a diversified, wide-moat business.

Big picture: Amid immense supply constraints across tech hardware, MPS stands apart with unconstrained supply. We like management's investments to increase its capacity past $6 billion in revenue, supporting growth ambitions for the next couple of years.

Our fair value estimate is $1,650 per share. Our valuation implies a 2026 price/adjusted earnings multiple of 60 times and a 2026 enterprise value/sales multiple of 18 times. Our valuation implies a multiple of 44 times against our 2027 non-GAAP earnings estimate and 35 times against our 2028 non-GAAP earnings forecast.

We forecast 27% compound annual revenue growth through 2030. We expect the firm to take share and win new sockets during this period across its served markets. We expect growth to taper down from more than 20% (and 50% in our model in 2026) to the mid-teens by 2030. Long-term, we expect the strongest growth for MPS in the automotive and data center markets. In automotive, the firm is quickly moving up the value chain into advanced driver-assist systems and onboard electric vehicle systems. In data centers, MPS benefits from strong placement alongside GPUs and custom accelerators for generative AI models, server CPUs, and networking chips and optics as well. Eventually, we think MPS’ market outperformance will moderate as its top line swells and market share gains come more slowly, but we still expect strong, above-market growth through the decade.

Over the next five years, we anticipate gross margin holding in the mid-50s with a rising mix of data center sales that keep margins at the low end of management’s long-term range of 55%-60%. Longer-term, we think that a higher mix of automotive revenue, along with greater volume and tapering data center growth, will push non-GAAP gross margin toward 60% by 2035, from 56% in 2025. We like the way MPS manages gross margin, continuing to invest in new growth opportunities that sometimes carry a slightly lower margin.

We think operating margins have room to expand with revenue growth but expect organic investments to stay high to fuel growth. MPS is expanding its manufacturing capacity and diversifying its geographic footprint as demand heightens and it grows larger. We think management will keep reinvesting in the business as long as it can realize continued outperformance of its underlying markets, which we expect will continue at a similar pace for several years. We expect non-GAAP operating margins to surpass 40% by 2030, compared with 35% in 2025. Management targets rising non-GAAP operating expenses at 80%-90% of the rate of revenue growth, but admits this isn’t as achievable when revenue growth is above 20%. We model the firm hitting this target in the outer years of our forecast with lower growth, but see stronger margin expansion in the medium term.

We think Monolithic Power Systems boasts a wide economic moat, as a result of intangible assets in power chip design and switching costs for its integrated power management chips. In our view, these will enable the firm to average returns on invested capital in excess of its weighted average cost of capital, more likely than not, over the next 20 years.

MPS' value proposition is eponymous: it can provide a single (monolithic) integrated power chip to replace a system otherwise composed of several discrete transistors or chips. MPS core competency is in voltage regulation chips, which control power to provide a precise voltage into a compute chip, whether it’s a GPU in a data center or a more basic processor in industrial or networking equipment.

For power chip customers, higher performance means better power density and a smaller form factor. These combine to offer lower power use, saving cost and reducing cooling and ventilation needs. We see the firm being consistently ahead of peers in terms of power efficiency and performance, which we attribute to MPS’ proprietary bipolar-CMOS-DMOS process technology. BCD combines analog, digital, and memory components into a single integrated chip; the combination of these three components creates a final product with higher power density and configurability than a combination of three point solutions. While other power chip players also have BCD technology (STMicroelectronics invented it in the 1980s), MPS has made it the core of its entire business, and uses it as an innovation vector to remain on the forefront of power chip design.

MPS also offers a uniquely capital-light model in power chipmaking, which enables immense returns on invested capital and a structural advantage over its competitors. Analog and power chips generally sell on the basis of reliability rather than cutting-edge performance, and large analog chipmakers typically build their own manufacturing sites (fabs) with lagging-edge manufacturing processes and extract high returns from producing the same product over the course of decades without much additional development. MPS has opted instead for a fabless model, designing its own manufacturing process (BCD) in tandem with third-party foundry partners. MPS uses its foundry partners’ older, fully depreciated fabs to design its own production lines and tools for its proprietary BCD at a relatively lower cost. By not owning its own fabs, MPS can be nimbler in innovating its process technology than larger peers that are content to produce lagging-edge, steadfastly reliable chips that produce steady cash flows. We believe this has allowed MPS to progress to higher-performance power chip designs faster than competitors.

The combination of unique design and a fabless model allows MPS to consistently remain at the leading edge of power chip performance and gain meaningful share from incumbents. Since 2020, MPS has increased its share of the voltage regulator chip market from 5% to 11% (2025) as a result. We expect these share gains to continue over time, with MPS enjoying share gains across end markets.

We think MPS' tech would be difficult for another established analog chipmaker to replicate, much less a new entrant. In fact, management has indicated that few in the firm itself actually know the process end to end, in order to preserve the firm’s secret sauce. In our view, talent for analog chip design is scarcer than it is for digital chips, due to greater emphasis in the broader semiconductor industry being placed on digital improvements for processors in smartphones and AI infrastructure. Analog chip designs are highly proprietary, and it can take years to train new engineers on the intricacies of analog design. We think that existing analog competitors are unlikely to steal enough scarce engineering talent to develop an entirely new line of products that could overtake MPS' technology.

We also think MPS benefits from customer switching costs for its power chips, resulting from its high levels of integration and low relative cost. MPS' highly integrated chips allow its customers to create unique, streamlined designs that couldn’t be achieved with competing offerings. MPS can offer a single chip that can take on the task of as many as six chips from its rivals. If a customer wanted to switch suppliers, it would have to undergo a complete system redesign, likely incorporating several different chips to achieve the same functionality that a singular MPS solution offered.

MPS' switching costs are steepest in its longer-cycle businesses, specifically the automotive and industrial verticals. In industrial applications like motor drives, end products can have lifecycles of a decade or more. In automotive applications, MPS works with OEMs or Tier 1 suppliers two to four years in advance of a car hitting the market, and model lives typically range from five to seven years, creating a total relationship length of up to 10 years. A customer is unlikely to want to switch suppliers mid-product cycle, which would require a system redesign and halt production for the duration of the switch.

In our view, the combination of MPS' differentiated process technology and the switching costs posed by its integrated chips carves out a wide economic moat. Due to its fabless model requiring low capital expenditures compared with other chipmakers, MPS is able to earn high returns on a small invested capital base. By our assessment, MPS’ returns on invested capital haven’t dipped below 20%—even when its business was barely breaking $100 million in revenue—and have steadily grown with its top-line and margin expansion. Over the next 20 years, we think MPS' asset-light business model will allow it to continue to earn returns above its cost of capital over the course of the business cycle and through heightened competition as it grows.

We believe MPS has the most rapid pace of innovation among power chipmakers, leading to fast time to market, design wins, and market share gains across its diverse end markets.

MPS has strong power management placements alongside GPU and custom AI accelerator vendors, providing an immense growth opportunity from generative AI investment.

The firm's fabless model is unique among power competitors and allows for terrific profitability and returns on invested capital.

MPS holds a small share of the power management chip market and will be challenged to steal share from large, moat-endowed, well-established competitors.

If larger analog peers like Texas Instruments or Analog Devices committed resources to matching MPS' process technology, this could hamper the latter firm’s growth.

The majority of MPS’ sales come from markets with relatively short product cycles. Stickier long-cycle markets like automotive and industrial still only make up a minority of the top line.