Carburton Street, London -- Years ago, my mother told me a story about the man who invented "cat's eyes". Apparently, he placed wine bottles along each side of his driveway to help him navigate home with reduced headlights during the Blitz. My mother told me that he now gets 1p for every single cat's eye installed in this country. I have no idea whether this story is true (I suspect not) but I remember being stunned at how amazingly rich this man must be. ARM Holdings(LSE: ARM) could quite easily find itself in this position with respect to the chip market.
ARM was established as a joint venture in 1990 to develop processor chips based on reduced instruction set computing (RISC). RISC works by packaging together the most common series of processor instructions. These make up a disproportionate amount of a processor's workload and, by concentrating on them, RISC chips are highly efficient, operating at lower power and generating less heat than other types of processor.
ARM is not just the market leader in RISC chips, it virtually is the market. When I first came across the company five or six years ago, everyone was naturally very excited about the technology. Some people were even suggesting then that, in future, most embedded chips would be designed this way. However, the perceived wisdom at the time was that they would just get squished by the big American chip manufacturers. As a result, most people (including myself) left the company well alone.
Well, we missed out. The company didn't get squished. Far from it. By deciding to license the designs to anybody that wanted to manufacture the chips, it made it so that it wasn't worth anybody's while to squish them. Companies like Alcatel, Cirrus, Fujitsu, IBM, Intel, NEC, Panasonic, Philips, Samsung, Sharp, Sony, Texas Instruments, Toshiba and Yamaha (pretty much every chip manufacturer worth its salt) pay ARM an upfront licence fee for using its designs and then a small royalty (averaging about 7p) per chip sold. On top of this, ARM makes revenues from consultancy, support services and selling development systems for its products. Royalties essentially feed directly though to profit, so the business model has extremely high margins.
Someone might come up with better designs but, then again, they probably won't. Remember that the ARM design is right down at the "putting instructions together in a logical and efficient way" end of things. However you build a processor, whether it's made from silicon, gallium arsenide, biological material or chocolate, a good chip still needs to process instructions in the most efficient manner possible. This is what ARM sorts out and it sorts it out far better than anyone else. Its business model also gives it the resources for research and development to make sure that it keeps itself ahead of the game. In any event, even if an alternative does appear, it's not as if it will land on the market tomorrow. ARM has been developing its solutions for ten years and it's only just getting started.
Applications
Anywhere a little computing is required, power is limited and there is little scope to dissipate heat, an ARM designed chip will most likely suit. This covers a huge range of applications. Mobile phones, hand-held computers, digital cameras, video recorders, hi-fi, microwaves, computer network devices, entry systems, baby alarms, central heating controls, talking toasters (I've never quite worked out if this is a joke), smart toothbrushes (this might be a joke too)... the list is pretty much as long as your imagination.
Cars are expected to be awash with basic chips in the coming years. Potential applications include braking systems, gear changing, airbags, climate control, keyless entry and customisation, navigation, hi-fi... Again, the limiting factor is your imagination. In a few years' time, it is thought that a low-end car will have around ten chips on board, while a high-end model might have as many as fifty. And it's not just cars, I could use a few chips on my bicycle.
Another potentially vast area for ARM is smartcards and "Bluetooth". Credit cards, bank cards, ID cards and any other card that you can think of (yes, maybe even birthday cards) will probably incorporate a chip in future. These chips are likely to be pretty basic and ARM's royalty will probably be pretty small, but there could be an awful lot of them about. Bluetooth chips are being developed as a means by which electronic devices can communicate with each other at short range, without the need for a physical connection. Imagine your TV, video recorder, set-top box, hi-fi and PC all linked and working together, without a single wire connecting them. Fantastic! You're also likely to find a Bluetooth chip (on top of the more sophisticated processor that's already there) in handheld computers, so that they can use the phone that's in your other pocket to access the Internet or send an e-mail. In fact, it probably doesn't stretch things too far to think that most pieces of electronic equipment might eventually contain a Bluetooth chip.
A Great Company on a Great Valuation
So, ARM is a great company. No one really disputes this. What people have a problem with is its valuation. On traditional measures, ARM undoubtedly has a rich valuation. But it is not a traditional company. In the past, companies have tended to have more predictable prospects. Valuation came down to finding a central case scenario and buying the shares if they were cheaper than this. These days, with the fantastic pace of technological change, this approach is becoming less useful and, for companies with highly uncertain prospects like ARM, it simply isn't valid. To show this, I will have to stick my finger in the air and do some hard-core guesswork.
The following table sets out five possible scenarios for ARM in 2010. Scenario 1 is the most optimistic and Scenario 5 is the most pessimistic. In each case, I have assumed that there will be a potential market of 2 billion people for ARM's products. The higher-end products might only reach 1 billion people and the lower-end products might reach 3 billion or more, but 2 billion across the board seems fair. Chips means the chips that might go into the applications that I outlined above and "replacement rate" means how often the average chip is replaced. Oh, and by the way, I really did do this on the back of an envelope.
Scenario 1 2 3 4 5
Chips per person 50 40 30 20 10
Total chips ('m) 100,000 80,000 60,000 40,000 20,000
Replacement rate (yrs) 2 3 4 5 6
Total new chips pa ('m) 50,000 26,667 15,000 8,000 3,333
ARM Market share 90% 80% 70% 50% 30%
Total ARM chips pa ('m) 45,000 21,334 10,500 4,000 1,000
Royalty 11p 9p 7p 5p 3p
Total r'ty revenue (£'m) 4,950 1,920 735 200 30
Other rev'ue (% of r'ties) 50% 45% 40% 35% 30%
Total rev'ue (£'m) 7,425 2,784 1,029 270 39
Post-tax margin (%) 50% 45% 40% 35% 30%
Post-tax profit (£'m) 3,713 1,253 412 95 12
2010 P/E ratio 80 60 40 20 10
2010 market cap (£'m) 297,000 75,169 16,464 1,890 117
2010 share price £301 £76.2 £16.7 192p 12p
You might say that a 2010 ARM valuation of £297 billion is unlikely, and it is, but bear in mind that Intel is already valued at about £275 billion. Is it really too much to think that ARM might, in our most optimistic scenario, be worth a similar amount in ten years' time? You might also say that the most pessimistic outlook of 12p per share is beyond the pale, but let's not forget all the wonder shares which have gone badly wrong.
Many people would say that, since £16.70 is our central case share price in 2010, we should use this as the basis for our valuation. We would compare it with today's share price of £6.50 and work out that we could get, as a central case, an annual return of 10% until then. This is hardly going to set the world on fire, so you might say that the shares are fully valued. However, to look at it this way is misleading.
Instead, let's reckon that there is a 20% chance of each of the scenarios occurring. The weighted average share price in 2010 is £79.19. This is our statistical expectation of the share price in 2010 and it is this which should form the basis of our valuation. We can now see that our expected annual return is around 28.4% -- very nice indeed.
Even if we decided that there was only a 5% chance of the most optimistic scenario (number one) occurring, 10% for scenario 2, 20% for 3, 30% for 4 and 35% for 5 -- a very pessimistic outlook in my opinion -- then our expected share price in 2010 (that is, the weighted average) would still be as much as £26.63. This would imply an annual return of 15.1%: plenty to be going along with.
The point of all this is that with highly uncertain situations, people are often so blinded by the possible downside, that they fail to take proper account of the potential upside. This approach is very misleading and is precisely what causes companies like ARM to be consistently undervalued. No one is disputing that an investment in ARM might lose you money (not me anyway) and you certainly wouldn't want to have it as your only shareholding. However, as part of a balanced portfolio, it represents an excellent proposition. The question is, are you brave enough to play the percentages?
Format for printing | 
Archive | 
Have your say
