Should you invest in ARM Holdings?
Or is it a bit overpriced?
ARM Holdings (ARM) is a great company, and has a fundamental role in the chip industry as a leading design firm essential to key players like Apple (AAPL) and NVidia (NVDA). ARM revenue comprises licensing fees, royalties and fees for tehcnical support. It is capital light, and earns a high return on sales and on capital employed. What is not to like?
The market price of the stock is a good place to begin. ARM is growing. The company went public only seven years ago but had history before going public, and with current sales of about US$2.6 billion, has a historical growth rate of royalty income in the 18% range.
Total revenue grew just over 35% in the last full year before the 2023 IPO.
With 1.03 billion shares outstanding, ARM is valued at just over US$50 billion which is a multiple of over 70 times the company’s EBITDA of just under $700 million. ARM trades at about 20 times sales and roughly 100 times earnings.
If you have a ten year investment horizon, and ARM can keep growing at 35% for a decade, by 2033 ARM will have revenue of ~US$45 billion and net income in the range of US$10 billion. ARM’s customers aren’t growing at 35% a year, the economy isn’t growing at 35% a year, and the industry isn’t either. A decade of 35% a year compound growth is not impossible but it is certainly unlikely. If it happens, and at the end of the decade ARM trades at a market multiple in the 20 times earnings range ARM would have a 2033 market capitalization of US$200 billion and investors would have a compound rate of return over the decade of about 14%. Certainly possible but certainly speculative as well.
ARM went public on the craze over AI, as NVidia has. The idea that AI will drive significant demand for chips seems overblown. Sure, newer and more powerful chips will be designed (when haven’t they?) and the semiconductor industry should enjoy sustained growth, but paying up front for the right hand tail of a wide probability distribution up front rarely pays off for investors. The theory that AI will require markedly faster processors based on new designs is a bit dramatic - large language models are storage intensive more than CPU intensive and current CPU’s are already turning in impressive power metrics. A typical Azure vCPU runs at 1,000 MIPS and the Azure virtual machine has up to 24 vCPU’s. Higher processor demand will benefit fabs like TSMC and INTEL more than ARM in my opinion and ARM is not the only source of chip design.
There is little doubt that ARM will grow, but substantial doubt that an investment at $50 a share today will fuel your retirement. If Apple or NVidia turn more to in-house processor design, and need less help from ARM, it won’t end well for those a bit too bullish today. That seems low risk, given NVidia tried to buy ARM and Apple just inked a deal with ARM that lasts to 2040 or later. That is old news, already in the stock price. Who knows, but ARM is too rich for my blood and its return to IPO prices after a short spurt higher suggests I am not alone in my skepticism. I like the company and avoid its shares at today’s prices. If the shares fall out of favor and drop significantly I might become a shareholder of ARM.