Arm Holdings is a company which licenses chip designs to semiconductor companies in many different industries. The key to understanding this company is that they don’t make the chips themselves but license out design tools which results in a 95%+ gross margin.
I have followed this company for one quarter but had thought the growth catalysts were not large enough. However, there have been a number of recent developments which have changed my mind and I have started a position in ARM for some of the following reasons,
Revenue is growing at 47% this last quarter
Licensing is up 60% yoy, and royalty 37% yoy
RPO +45% yoy
96% gross margin last quarter
Full year adj operating profit of 1,408M +80% yoy
Q4 69% Revenue to cash conversion, 637M in adj FCF
The platform version v9 gets roughly double the fees as did the v8 platform
Say they are oversubscribed compute subsystems and have far more demand for the product than anticipated
I recently learned most Supermicro racks that are loaded with 72 Nvidia Hoppers and also have two Intel x86 CPUs. The full rack solution still requires CPUs on board
The Nvidia Grace CPU will be competing to take the place of these Intel chips and the main selling point is they are twice as energy efficient, although they offer the “same” performance as x86 chips
Their guidance projects out weakness primarily because markets like phones, automotive, and IOT had seen some slowness
AWS is also building a chip off Arm called Nitro for storage offload
Able to work in China through Arm China
In their own words, “The AI tailwind, has driven unprecedented growth for our licensing business”
GPT and Meta’s Llama rely on the Arm compute platform
Can “accelerate the total development cycle of an automotive solution by up to two years”
Enable chips makers to develop Arm based chips at the Intel Foundry (ARM base chips can be made at Intel fab)
83% of their employees are engineers!
Strong leadership team which promotes from within (Note that company is UK based but the CEO is in Silicon Valley)
CEO used to work at Nvidia for 7 years as vice president and general manager of its computing products business
CFO was also CFO at Splunk prior to this
An engineering culture which promotes from within. Here’s the description of the head of engineering, “He joined Arm in 1998 as a CPU designer, eventually progressing to engineering leadership as Vice President of Engineering for Media and CPU.”
One important item to note is that this last quarter benefitted 166M in net income because of adjustments related to stock options, so the EBITDA numbers may be a bit more useful than the net income numbers.
Potential drawbacks that I see to investing in this company are there are a lot of legacy aspects to the business of supporting older versions of the software and clients in IOT. Here’s what they say about the IOT market, “IOT chips are very high volume, but relatively low value”. I want to make sure the company stays focused on taking the AI market and hope that these flat parts of the business don’t drain the business too much.
Curious for other’s takes on this company? Interested to hear if the 95% gross margins makes this more appealing than owning other semiconductor companies?
I used to hold ARM for many years and it was my first 20 bagger. I have a soft spot in my heart for ARM as they were the ultimate little company that could.
When it was brought back to the public market I didn’t buy into the story as I saw revenue as being flat to single digit growth but that was entirely driven by price increases and actually masking volume declines. This doesn’t correlate with the revenue growth picture this latest quarter is suggesting so maybe I have to go back and re-evaluate. Or maybe they have raised prices again.
Whatever the story is I urge you to look at not just revenues with ARM as they are a very different model to other semiconductor companies as they are a fabless IP company.
Firstly you have to look at license vs royalties. Licence revenues are one off, royalties are indefinite.
Secondly you have to track the underlying volume growth rather than top line revenue growth as that represents the real state of the business.
Third you have to look at the cents/unit as this is important to see whether their innovation is producing more value or whether they are achieving volumes in low value commoditised sections of the market place.
Will go back and take a look to see if there is more to this story when I can.
Couple things…. They are not using Intel as their foundry. They provide their cores as intellectual property, and ported to various fab processes that their customers may want to use. Intel is thus one of many. I currently use UMC for example. Every node has various ARM cpu cores available and each customer pays some form of recurring and non-recurring fee.
As Ant states, because the deliverable is IP, the gross margins should be high as there is no product cost, no hosting cost, etc. There is some latitude in accounting to account for:
Employee costs for customer success / support / customer service
Customer onboarding costs
Software companies often put these in the product cost. There might be some of this being put in SG&A and thus showing a higher GPM.
Is Softbank’s 90% ownership of ARM still having an impact? The float is relatively small and SoftBank might want to take profits at any time.
The forward P/E for ARM was pretty high- what is it today? If you’re going to cite the Grace chip so much, wouldn’t Nvidia profit more, like Apple with its ARM chips?
o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o
This might not be the “new normal” but the result of the recent breakout of AI compute. It could be a value trap if this happens to be the “Peak of Inflated Expectations.”
ARM might have already passed that peak (chart since IPO)
o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o
SoftBank is the hard to predict 800 pound gorilla in the room. They could crash the stock price at any time making life difficult for investors but I doubt they have long term effects on the stock.
Thanks for all the feedback so far. I wasn’t aware that SoftBank still owns 90% of the shares and that lowers my confidence some. Sounds like the company was acquired for 32B in 2016 and then spun up again for an IPO in 2023.
For those that know about how the royalty rates work I’m trying to understand how much ARM would make per Nvidia CPU. I’ve read that ARM typically charges between 1-2% as their royalty fee and that the Nvidia Grace CPU Superchip integrates 144 Arm Neoverse V2 cores.
Would that mean that Nvidia pays a single 1-2% fee per Nvidia chip? Does it matter how many Arm cores there are in the chip, or just that Arm is there at least once? I believe their formula for royalty revenue is basically,
Royalty Revenue = Selling price of chip * royalty rate
The forward P/E for ARM was pretty high- what is it today? If you’re going to cite the Grace chip so much, wouldn’t Nvidia profit more, like Apple with its ARM chips?
It’s still a pretty high P/E trailing 12 months is listed as 400. Not sure on the forward P/E though but my bet here is that they will be able to get more operating leverage as they go.
I have an investment in Nvidia too, but it seems like a pretty good business that Arm can take 1-2% of each sale especially if the Nvidia CPU reaches mass scale.
This is the main reason I’m not interested in it. People focus on the 95% gross margin because they are not selling a physical object. But their Net Margin Profit is 9.5% vs Nvidia which has a Net Margin Profit of 48.8%. Nvidia is growing at over 100% to Arm growing at 20.7%. Nvidia’s operational expenses are growing at 22.2%, where Arm’s operational expenses(Majority of their expense) are growing at 58.6%. Nvidia P/E is 66.5 vs Arms 416.04. Why would anyone pay 6 times as much for Arm as Nvidia on P/E. Nvidia is growing sales faster and for each dollar of sales they pocket 4 times as much money as Arm.
Arm is a great story but I believe its way overvalued even for growth investors. The finances does not support its stock price at all.
That’s a good callout as the valuation is already stretched pretty thin, especially when you put it side by side with Nvidia.
I actually ended up closing my position already after reading into Arm China a bit more from the S-1 document. From the document one of the risks is, “Neither we nor SoftBank Group control the operations of Arm China, which operates independently of us”
It seems like Arm China is in a very grey area where they’ve basically gone rogue as some people put it. They still send revenues to Arm, but Arm has no control over Arm China. It turns out Masayoshi Son, head of Softbank had to get back control of Arm China to even IPO. This is a very complex relationship between Arm and Arm China to say the least.
My main reasons for selling this soon are,
Discovering SoftBank owns 90% - seems like a huge risk that the fate of Arm is tied to SoftBank
Valuation is stretched and has a lot of optimism priced in
Arm China is not even owned by Arm although it contributes some 20-25% of revenue
These last 2 points together with the absence of volume growth were the principle reasons why I didn’t jump on board the ARM train when it returned to the public market with the Softbank IPO and that’s saying something given that apart from Denny I probably had the longest invested history in ARM the first time around. Softbank’s holding doesn’t bother me so much.
As a consumer, I love ARM’s approach and business model. It took the wind out of Intel’s one-size-chip-fits-all approach and their huge resulting profit margins. We never would have had all the variety of CPUs we have today, many special/limited purpose, without them.
As an industry investor, though, I always felt ARM were leaving too much money on the table. Clearly, I was wrong on the stock appreciation potential side of things, but the buyouts and re-IPOs do somewhat make me feel better about my flawed analysis.
