ARM - Kerry McGuire, Director High Performance Processors
MIPS Technologies - Jack Browne, Vice President Marketing
Rambus - Michael Ching, Product Marketing Manager
Sonics - Phil Casini, Vice President Marketing
Tensilica - Steve Leibson, Technology Evangelist

There seem to be a great many IP vendors in the market now, and though arguably it is already overcrowded, still more seem to be emerging. At the same time it is no secret that probably only five or six players in this sector are actually profitable. Have too many people been tempted to found their own IP firms, as they perceive it as an easy option, and likewise are VCs too eager to offer money to start-ups in this space?

Phil (Sonics) - The term IP vendor is now such a broad term, each looking to address a specific area, and that is a sign of the maturity of the market. I don't think that people should talk about the IP sector in generalised terms anymore, it is more a matter looking at which area they look to address.

Steve (Tensilica) - I'm not sure that it is important whether or not it is too easy to get into the IP market. It's a bit like asking if the Internet has made it too easy for people to get into the publishing market. This had allowed more ideas to be realised, and helped produce products that would have otherwise remained on the drawing board.

Jack (MIPS) - With regard to the market being to easy to get into, I think that exactly the same thing can be said to be going on in any market. There will be an ongoing processing of new firms starting out, some will be able to establish themselves as long term players, others will just be one hit wonders.

Phil (Sonics) - There actually have to be quite significant investments involved, and although for simple things like producing an I/O interface the cost of entry is low, if you are addressing more significant processing or interconnect issues, the money needed to deliver a reliable long-term solution is going to be considerable. What we have found is the harder the problem you solve the greater the value you can bring to the market.

Is their going to be further consolidation here as the sector matures, or can it support all these firms?

Steve (Tensilica) - I would say this will be a cyclic process, new companies emerging, some being bought up, others making it for themselves. There will always need to be an element of sorting through the rubbish. The barriers to entry in the semiconductor industry were lower in the 50s and 60s, so plenty of firms sprung up. It was the consistency and reliability of certain firms that allowed them to differentiate themselves from the rest of the pasck.

Kerry (ARM) - I don't see any huge kull occuring, there will always be casualties, but no more than in other sectors in the electronics industry. Longevity requires investment, not just luck in getting something right at one point. It is being able to offer proper support for youir products and partnering with major firms that really counts.

Are royalties more important than simply getting licenses? Do we have enough new designs taking place to support this, or is it more a question of aligning with a couple of big player that can offer high volume output?

Phil (Sonics) - It is reallly a matter of what value you are bringing to the market. If you are consistantly improving what you can offer all the time then you can wade out that period, but if your value dimishes over time, then you will never reach the point where you get beyond the start up phase, and any potential you had will be lost. It is the innovative players that will survive.

Jack (MIPS) - Our philiosphy is that if we have a royalty element to our business model then we can make the licence fee lower, and this decreases the barrier to entry for the customer. This helps them, as they don't have a big upfront investment. A lot of chip designs don't get to first silicon so having both components parts is important. Maybe one in four designs get there. Cashflow is more important than profitability for a start up, so yes it is a tough to get through these leans years, but it is those that do who will reap the benefits.

Michael (Rambus) - Yeah you really need to focus on innovation. Companies will need to go through a period of austerity, and they will have to have VCs with deep pockets and a strong bvelief in what they are doing behind them, but those who have something to offer will make it through.

Would you say companies like ARM and MIPS it was a matter of good timing; both of you being around at the point when a new market opened up (e.g. the mobile market in the case of ARM and the games console market in the case of MIPS)?

Kerry (ARM) - I think that it has not been simply the timeliness of our offering hitting the market that really defined our success. There is far more to it than that. It has been thanks to our reinvestment back into the business that we have ensured our sustainability.

Does the fact that companies have go through several years of austerity before they can get any serious royalties, and have to rely purely on money from licences, limit their ability to grow and be innovative. Will this make it hard for new IP firms to break into the big time?

Steve (Tensilica) - But surely, that is true of any market? I think that everytime there is a major upheaval in the market there has been a new generation of companies that come to the fore, and firms that didn't make the transistion get left behind.

Phil (Sonics) - Yeah I would agree with that. For example, a few years ago who would have predicted that Google would be challenging Microsoft. I don't see any glass ceiling effect, those who make the right descicions will supplant those who become complacent.

Are we likely to see some sort of two-tiered structure, with a few big players offering a broad portfolio of IP solutions, and specialised vendors serving particular niche markets?

Steve (Tensilica) - Well nobody is likely to start buying up big chunks of this market, because integrating the other company's IP into their product offering would just be to difficult to do.

Kerry (ARM) - Acquisition is an option, but it really has to be complementary technology, not purely for size benefits.

From a processor perspective what are the main drivers as we move down to 65nm and beyond; saving space so more cores can be placed on the chip, reducing static power, or just plain number crunching?

Steve (Tensilica) - I would say three things, in order; power, power, and power.

Jack (MIPS) - Well basically the main factors are likely to be price, performance and power, but now programmability is also becoming an important concern, and this is going to become more so as the amount of software needed within a design increases.

Kerry (ARM) - As well as those already mentioned, the reliability, and robustness of the solution are important issuse here.

Steve, Tensilica moved from purely offering configurable IP to also having a standard IP portfolio about 18 months ago. How has that panned out? Have your diamond cores been embraced by the market, and do you feel this two pronged attack is beneficial?

Steve (Tensilica) - The market's response to our Diamond cores has been enthusiastic. That's what we expected; we created these cores based on feedback from our customers and prospects. Many SOC designers have control applications that are well-suited to a standard RISC processor core, which Tensilica could easily create using our processor-generator technology. Some designers needed a processor core that could run Linux, again something we could easily achieve with our processor-generator technology. Both types of cores can be found in the software-compatible Diamond processor core family. Finally, many SOC design teams had a common need for an audio processor, so again we've used our technology to produce the 330HiFi processor, which incorporates some 300 specialised audio instructions that allow the processor to execute a wide range of audio codecs at very low, energy-saving clock rates. In some cases, once a design team becomes familiar with the Diamond processor architecture and the associated software-development tools, some members of that design team have felt ready to use a software-compatible, configurable processor core based on the same architecture and they step up to a configurable core to meet performance or energy-conservation goals beyond the reach of any 32bit, fixed-ISA processor core. In other words, they might have called us about a Diamond core, but once they learned about Xtensa configurable processors, they switched.

Is the somewhat aggressive stance that Rambus has taken, regularly suing companies, proving to be problem? Isn't it hard to approach a firm for business after you have taken them to court?

Michael (Rambus) - What we are really focussing on is the technology. Protecting IP is certainly a factor, but we don't want this to overshadow what we are doing elsewhere. I think this is a startegy that has been well received by the market.

Phil, Sonics has established itself in a relatively new part of the IP space, namely the interface side. Do you see it becoming a more important sector in the future?

Phil (Sonics) - With the number of cores increases the data flow is going to become a vital issue, and the economies involved will mean that chip manufacturers won't be able to make this feasable in house. So yes I think the I/O side will become an increasingly important part of the IP business.

Some of the major fabs have invested heavily in developing their own IP in house, is this likely to prove a threat? At the moment they are obviously useful partners, but could they start to be competitors looking to steal your action in the future?

Phil (Sonics) - The larger semiconductor firms are developing some IP of their own, but it is difficult to rationalise two business models that have revenue streams that are so different. So I don't see them using this to steal our business.

Michael (Rambus) - We really see this sort of IP as being complementary. Primarily these firms want to sell wafers. It is in their interests to have some in house IP to help their manufacturing, but they won't want to get bogged down in the complexities of turning this into a business.

Kerry (ARM) - I largely agree with this. From the fab perspective the investment and the sort of resources they have at their disposal mean it wouldn't really make sense for them to pursue this.

Steve (Tensilica) - On the other hand companies like NEC and Fujitsu have developed a hybrid business model, in which they offer IP as part of their ASIC portfolio. So they are both licensers and licencess of IP. I guess you would say there is a continuum rather than an either or.

Jack (MIPS) - The return on investment models are very different, so a semiconductor company has a challenge here. Also there are issues about how you would offer innovation too your rival companies and still compete with them as a manufacturer. Thre are many conflicts of interests to consider.

How do you look to keep your IP resources safe without encroaching on your ability to sell your wares?

Kerry (ARM) - If it is harder to go down the piracy route, not just from a legal but a technical point of view then the reasons todo this become less clear.

Steve (Tensilica) - We have an interesting perspective here as our intelectual propoerty and ours is fused, together, so the customer will look to ensure that both ours and their interests are served. This gives us some added assurance.

Jack (MIPS) - As companies develop their own intelectual property, the more it respects the intellectual property of the companies it works with. When we engage with a company we start with hard core solutions, then over time we progress to giving them soft IP as the relationship builds. This way we are not over exposed to the threat of piracy.

Because some of the players out there are just not able to deliver what they promised, is it bringing the industry into disrepute? Is there a need for better guidelines? Do VSIA's standards get enough respect from the industry? Are companies taking it seriously, or are they largely ignored?

Kerry (ARM) - We see an investment in standards being very important, and being involved in these helps facilitate our business model. The possibility to offer ‘one-stop-shop' is suitable for some customers, but it grows the market as a whole.

Jack (MIPS) - There should be a set of standards to keep this market in check, but though you can meet all the requiremnts of bodies like OCP-IP and VSIA you might miss out important requirements from your cuistomers. So it is true standards have their place, but it is not the whole story.

Given the complexity of SoC designs moving forward will mean that pieces of IP from various vendors will need to be integrated together, how big a headache is that likely to prove, and does that invite great reliance on standards such as those of the VSIA, or will it favour larger companies who can offer more of the components IP parts?

Phil (Sonics) - It is really important to drive interoperabilty, with IP libraries getting more complex, the porting costs have to be taken in to account, not to mention the problems of upgrading older IP based on a different processing environments. So being able to unify these things is a great benefit.

Where are the new IP opportunities going to come from; memory, interface, processing, medical, portable consumer?

Phil (Sonics) - The convergence between communication and multimedia isn't going to stop, so I think there will be great opportunities in HD decoding

Michael (Rambus) - I think that it will be solving the signalling issues that we will see as we move down to lower architures and start employing higher speeds.

Kerry (ARM) - Graphics processing is clearly going to be a big growth market, given that demand to bring TV into the mobile arena. In the longer term it is likely that areas like implantable electronics will show great potential.
...............................................................................................................