Bangalore: Tudor Brown, president and co-founder of UK-based chip designer ARM Holdings Plc, was in Bangalore overseeing the induction of new hires in the company’s 350-strong research and development centre in the city.
Sharing experience: Brown says it is a huge commitment to make a chip, and if firms are trying to do that on the cheap, it will never work. Photo: Aniruddha Chowdhury/Mint
In an interview, Brown spoke about his vision for ARM and what an “under-capitalized” India needs to do if it wants to become successful in chip design. “India frustrates me,” he says in a well-intentioned remark. Edited excerpts:
Are you reorienting what you are doing in India?
Not really. We effectively started in 2004. My initial vision was to create a microcosm of all engineering for ARM. But you need critical mass. So today we have consolidated around two major activities. One is the physical IP (intellectual property) side, which continues to grow. That is half the people. The other half is around validation for processor design and architecture.
Do you think India as a country has moved up in its capabilities for high-end chip design? We are even talking about a national fab (chip fabrication plant).
India frustrates me. I see a lot of enthusiasm, but I don’t see much execution, I am sorry to say. Fundamentally, it is under capitalized. It is always trying to do things on the cheap. If I compare India and China, and I go to both quite a lot, you could argue they both have come from equivalent positions in terms of gross domestic product or whatever. Yet, Chinese companies have raised the capital to buy not just ARM licences, but also the whole ecosystem needed in terms of electronic design automation (EDA) tools to build chips and get products out there. Indian companies are still talking about it. But whenever we talk to Indian companies, the fundamental assertion is “we have got no money, so we need a cheap licence”. If you want to play on the world stage, you have to play by the world rules. Of course, costs are low in India, so you should be able to do things at lower cost. But you can’t expect companies like us, EDA companies and international suppliers to single India out as a special case. Then what about China and Brazil, which are doing things pretty much at full value?
The other aspect of this is that there needs to be a lot of skin in the game. It is a huge commitment to make a chip—in terms of money, resources, the doggedness to see it through, debug it, revise it, etc. If companies are trying to do that on the cheap and not put skin in the game, it will never work.
As for a national fab, I don’t understand why India needs one. Many fabs around the world have demonstrated that you need critical mass and you need to be leading edge. Even in China, they have invested enormous money into fabs. They are not leading edge and are struggling to make a go of it. And if India wants a fab that is not leading edge, it would be much cheaper to buy capacity in pre-existing fabs that have been amortized.
If you are talking about national requirements, defence, etc., that only works if you can get to a leadership position. Whether it is armaments chips or high-volume commercial chips, the fact is there aren’t many of these products that are built on leading edge, third generation fabs. I would respectfully suggest that resources are much better spent on designing and building products.
Sure, you have these enormous requirements and you have an import problem. However, the value is not in the chips, but in the end products. So that is where the investment should be made.
Where do you see ARM going in its drive for market share in spaces other than smartphones?
I am absolutely convinced that the base business will expand and we will gain market share, outstripping growth in the market, covering an ever-wider spectrum. The “Internet of things” hasn’t started yet, but we are absolutely convinced that ARM will be in all those things once it starts.
ARM’s sweet spot in mobiles will expand, so will the high-volume lower end with microcontrollers and so on. As will the high-end, low-volume market in servers, where the value is significant. We have said that we should have server chips in the market by 2014-15 in significant volumes.
What technology trends do you see?
One is towards multi-core designs. The other big area is graphics. Arguably we have the leading set of graphics cores in the embedded space at the moment, in terms of technology if not volume. We want to expand that. So you have multi-core central processing units and multi-core graphics processing units. The challenge is to make them work together efficiently. We have the technology and that will be introduced in the next generation.
Take security, solutions for which have been in ARM designs for a decade now. Modern devices can do so much, the need for security has increased. We all survive with these ridiculous credit cards today that have almost no security. And the credit card companies and retail stores are spending millions of dollars on (combating) fraud every day. And they accept that as the cost of doing business.
Then there is lowering power consumption, another strength area for us. Phones are a fantastic driver of technology, if you can solve the problem of the phone you solve many others. Look at laptops. We are all fed up with the pathetic battery life. Even phones. Why should we charge them every day? When ARM laptops come out, you will see all this change.
When is that likely to happen?
They are already there, unless you are thinking about Windows on ARM architecture. Microsoft is very serious about it. I would like to think by the end of the year, but I really don’t know that. But when it arrives, it would be very interesting to see what happens. Intel may have this Ultrabook hype about instant-on and long battery life. But that is a big issue for them because it is difficult for Intel to do it. But with Windows on ARM, it will be easy to do it.
You are also targeting the server market, another Intel stronghold. Are you ready for the counter attack?
The ARM story is compelling in that space, no matter how hard Intel tries to fight. That is my belief, so we will see. It is a life and death space for them, though it is not for us.
sridhar.c@livemint.com
