International Electronics Forum, Geneva, Switzerland 1 and 2 October 2009

Citation

(2010), "International Electronics Forum, Geneva, Switzerland 1 and 2 October 2009", Circuit World, Vol. 36 No. 1. https://doi.org/10.1108/cw.2010.21736aac.002

International Electronics Forum, Geneva, Switzerland 1 and 2 October 2009

Article Type: Conferences and exhibitions From: Circuit World, Volume 36, Issue 1

Day 1. 1st October

As the leaves fall from the trees, so did the panoply of information at the International Electronics Forum in Geneva this autumn. This is the eighteenth conference run by Future Horizons, and the 100+delegates were welcomed by Malcolm Penn, Chairman, who commented on the changes that had taken place to the timing, the arrangements, and the location of the conference, as well as the industry as a whole during the year. It had been a lousy year, with a zero for several quarters, but such things do not last, they never do. Sunlight was to be seen, and not only outside on the mountains. Whilst Malcolm Penn was noting his 40th year in the semiconductor industry, one of the delegates was celebrating his 50th year, which means that he must have started before the transistor arrived, and steam trains were still running in much of Europe.

Daniel Loeffler from the Republic and State of Geneva puffed into open the conference. He spoke about the culture of communication, and how this will change our behaviour. The Economic Development Office, which he represents, oversees a dynamic business base in Geneva, but he is aware of the competition from outside the region. He remarked on how the watch industry was moving its expertise into the medical field, an example of how market forces had brought about a paradigm shift in traditional manufacturing, whilst utilising local unique skills.

Maria Marced is the President of TSMC Europe. Innovation can stimulate demand was her theme. We have seen the first negative global GDP since WW2, but the future, according to the IMF, looks better with a return to 2005 levels by 2011. Q4 and Q1 were disasters, but leading indexes show stablisation, with China at the forefront. Recovery follows every recession, and this recovery contains plenty of exciting new opportunities – “lots of silicon in my handbag” – as she so aptly put it. Where technology enables innovation can be seen in the consumer, computing, and handset fields, but emerging markets demand lower prices, and thus lower margins, at a time when R&D investment increases. Here, collaboration is a must; it leads to more reuse, and smaller total investment. At TSMC they are focussing on four pillars of investment – leading edge technology; mainstream technology; design technology and backend technology. They already have 1,200 R&D engineers, and 600 design technology engineers, and have now added 400 in R&D, and are hiring another 100 to design, globally. Continuous capacity expansion is the key, as illustrated by their new fab. plant in Taiwan, with 45,000 m² of clean room, a capacity of 1.8 million 12 in. wafers per annum, of which it is now running at 95-100 per cent. TSMC produce 11 million wafers annually, and are the one of the really sound fab companies.

Brian Harrison is the CEO of Numonyx BV, located close by on the shores of Lake Geneva. The world has an insatiable demand for memory, he announced, and his global company is meeting this, generating a turnover of $2.3 billion. Formed by a merger of Intel & ST, they are the third largest flash memory provider in the world, and the number one provider to the handset industry. There are two drivers for memory – the cost per bit, which has declined, although consumption has seen spectacular growth, driven largely by digital cameras, and more latterly iPods and MP3 players. MEMS were the other driver of growth. Numonyx are themselves driving growth on two fronts – with new technologies such as PCM (Phase Change Memory), which has better performance characteristics (and 200,000 years of memory) when compared to Flash and DRAM. To give an example of how fast PCM is (17 times faster than solid state) then the same amount of data on a PCM which would take a day to download, would take nine years from a HDD. Even better was the storage class memory (SCM) PCM, which will take 30 min. Thus, the SCM PCM will become a disruptive force, and with>150,000× access density improvement, they will need less hardware, less space, and use lower energy. Memory growth is driven by cost and capabilities, and Numonyx are delivering both. They have 45 nm technology PCM in development, and are prototyping. No one other than Numonyx and Samsung have working PCM. On the verge of something big, the memory market is going to go through another revolution, and we are looking at growth by a factor of 10.

The cheery Jean-Marc Chery is the CTO of ST Microelectronics – “More Moore” and “More than Moore” was the title of his presentation. Their 32/28 nm CMOS node presents a disruption, and has its origins in the IBM alliance with ST. Jean-Marc posed the question about getting the best supply path in the market, and this may not be done under one label alone; it may require an alliance of companies working together. Such an alliance is the ISDA, the International Semiconductor Development Alliance, which combines the companies such as Toshiba, Samsung, IBM, NEC, ST, Infineon and Chartered, which advances R&D through universities in the USA, Europe and Singapore. Their CMOS roadmap includes 3D-integration, the next big thing after 20 nm. 3D SiP in 2008, 3D WLP in 2009 and 3D SoC in 2011/2012. The IBM Alliance wafer sourcing plant at Fishkill in the USA is the development centre, from which 28 nm fabs will be ready for manufacturing by 2010 at four sites. Monsieur Chery showed pictures of the 9,000 m² fab linked to a 6,700 m² fab though a clean link in Crolles, near Grenoble in France, producing 250,000 20 nm wafers (5,000 wafers per week), Small series are a specialisation, and they offer medium volume complex-product production through a “smart” combination of internal and external sourcing. No end to “Moore”. Nor “more”, it seems.

Alain Dutheil is the President and CEO of ST Ericsson, a company created only in February of this year. They have a workforce of about 8,000 of which more than 86 per cent are in R&D, and they have sales of about $3.6 billion per annum. A fabless company, they produce platform and semiconductors for wireless, with their main operations in France and Sweden. Global semiconductor revenues between 2006 and 2009 were shown; they had declined in all areas, but less so in the handset sector; here the continual need for people to communicate had mitigated the downturn in wireless devices, which brought him onto broadband, the theme of his presentation. “For every 10 per cent points of growth in broadband penetration, the GDP of a country grows 1.2 per cent”, according to Philippe Dongier of the World Bank. In Ireland, GDP increased by $4 billion with the introduction of broadband. There are many examples of the contribution that broadband has made. Internet penetration rates have been at their best in the North America, followed by Oceania/Australia, then Europe, with Asia surprisingly down at 18.5 per cent, with a world average of 24.7 per cent. It is the mobile ‘phone that often enables the first internet access for the subscriber. Broadband changes to way we work and the way we live, and whilst Alain claims that it improves the quality of ones life, there are some who might dissent from this view. If you love statistics, then there will, by 2014, be 3 million people subscribing to mobile broadband, with over 550 million devices at the same time. The numbers are likely to change dramatically. The next step is LTE, or you may call it 4G, but it is the next broadband evolution, with incredible data speed: 100 Mbps downlink, and 50 Mbps up link. This will come to laptops first, then consumer electronics, and then smart ‘phones and MIDs. New consumers have grown up with the internet and now use it as standard practice, and rely upon the availability of data on line, and use social networks such as Facebook (with some 250 million users). Linking a TV to the internet, social networking, entertainment, location based services, gaming, advertising – these all drive the demand. LTE will kick-off next year, within smart” phones and connected devices. Just imagine for a moment 50 billion machines communicating amongst themselves worldwide!

Looking at another key industry, 3D stacking technology has many implications, as described by Nobuaki Miyakawa from Honda Research Institute in Japan. Set against the background of performance trends of semiconductor devices, we are now reaching the limits of performance with known technologies, so he expected technology for the future to be based upon 3D structures; wafer to wafer stacking technology based on 8 in. wafers (Thru Silicon Via (TSV)) chip sizes become small, and it is possible to realise high yield by means of small chip size, which means low cost and low resistance value by means of short wires. Here, connection between each chip using short vertical wires in between is made, performance is improved and power demand is reduced.

Connectivity between layer bumps is improved with fewer resistance steps, and fewer processing steps (down from 7 to 5), with a direct connection between TSV and bump. Honda has proposed a new fabrication method for a multi-layer stacking device using wafer-to-wafer stacking technology based on 8 in. wafers. The trial manufacture of three stacked layers using the proposed technology has demonstrated remarkable characteristics, and shown the trend of TSVs for practical use and possibility of better profitability using the stacking technology. The DTSV from Honda gave 100 per cent connectivity measured over 100,000 pieces, and test results on 8 in. wafers were illustrated in detail; the structures were highly reliable within the demands of MIL STD and JEDEC. When it comes to costs, using stacking technology will save $8,400 million per annum. When it comes to manufacture, yields are as yet a problem.

Cloud Computing was described by Young Sohn, CEO at Inphi, a fast growing semiconductor company in the USA. Powering this Cloud computing is anytime, anywhere computing, cheap storage, multicore processors and virtualisation under the technology banner, allied to energy efficiency, and of course, economics. Seven companies are involved with Cloud – Intuit, EMC², Google, Amazon Web Services, IBM, Windows Azure and SalesForce.com. With internet growth currently at 44 per cent pa., and with services such as YouTube, Facebook, Flickr, Hulu and Netflix being household names by now, Cloud is going to be $100 billion market by 2013. User benefits include access from anywhere instantly, and instant deployment and scale. It is energy efficient, offers performance, and security and reliability.

Consumption of energy by memory is 11 per cent, and as you increase the density of computing power so the advent of 4 GB of DRAM memory is now needed and at Inphi they have found a way of efficiently supplying servers with their power requirements, bringing about considerable savings from cooling. SSD controllers will improve performance hugely. We are all using Cloud computing to day if we did but know it, but energy efficiency, performance and security are the key challenges and opportunity for the semiconductor industry.

After an excellent lunch, Enrico Villa from Cluster for Application and Technology Research in Europe on NanoElectronics (CATRENE. spoke most eloquently about the key for global competitiveness. ICT. Of the key enabling technologies, one of them is microelectronics, and thus semi conductors. Electronic and information systems for society are worth $87 trillion and growing, about 10 per cent of global GDP.

As a result the semiconductor market is increasing at double the rate of GDP growth, and ICT patents account for more than 85 per cent of all high-tech patents in Europe, the USA and Japan.

For the future, there are opportunities for new technology as pressures increase on society's structures. For instance, 600 billion euros are spent annually on healthcare in Europe on an ageing population, and this is 60 per cent of the total spent on healthcare. It is unsustainable. Lighting accounts for 13 per cent of all energy costs, and 35 per cent of that could be saved by changing to new technology such as LEDs. Leadership in new markets is strongly linked to leadership in micro and nanoelectronics, and Enrico awaits a consistent European Industrial Innovation policy, given that semiconductors are a $248 billion industry worldwide, and Europe is a leading player in this field. Moving from ideas to products is one area where Europe is less strong, but he was happy to speak about the effectiveness of the Eureka project in linking European clusters which have brought about welcome co-operation between European players. Such cooperation has led to several European semiconductor companies being the very first on their market; two European equipment/material suppliers have emerged as world leaders, and Catrene has given access to the most advanced nanoelectronics technologies for European electronic systems in the fields of telecomm, aeronautics, automotive, defence, or industrial electronics. In answer to a question, he said that 5 billion Euros are spent on electronics R&D by the EC in Europe.

Ron Collett, President and CEO of Numetrics Management Systems Inc. was “raising the bar on semiconductor R&D Management execution and ROI”. Working with PRTM (a operational strategy consulting company) they have found that a profound change is taking place in the semiconductor industry. Companies are disintegrating, outsourcing manufacturing. There has been a sharp drop, 36 per cent in fact, of the numbers of people working in the industry between 2001 and 2009, reflecting the decline in manufacture. There are not the markets out there to support the ROI on the huge sums required for developing a SoC, which now costs 100s of millions of dollars, with an ROI measured over nine to ten years, for example. It is companies with well-differentiated products who survive, and it is companies who specialise, who have customer solutions, and who have differentiated products and product development superiority, who thrive.

To help people out of this pothole, PRTM have produced PACE^® as Industry Best Practice. Ron described this in some detail, how this system worked, based on excellence being achieved under several labels, functional excellence, project excellence, portfolio excellence and enterprise excellence. It is a fact that 60 per cent of all SC projects slip their completion dates by 25 per cent and 16 per cent of all IC projects slip more than one year. Fact-based project management allows for solid decisions to be taken. New Product Development success relies on non-trivial R&D basics in semiconductor design. Ron gave some examples of how various problems had been solved. A$1B+semiconductor company had a global R&D footprint but>40 per cent longer TTM than average industry performance. His company had Identified $26M in annual additional capacity by advancing to median benchmark performance. A multibillion-dollar global electronics company with $27M in R&D write-offs due to late-stage product cancellations sought their help, and Numetrics had demonstrated a 25 per cent reduction in R&D costs for the same R&D throughput. A $5B+electronic equipment company struggled to identify sources of new incremental revenue, and Numetrics showed a sustained 5 per cent incremental revenue uplift from strategic partnerships/alliance. Numetrics is a company who can ensure that their client's projects planned with their help exhibit significantly better on-time performance, and the case was made with great eloquence.

The enthusiastically entertaining Martin Orrell of TTP, based in Cambridge in the UK, had software applications on leading edge silicon as his presentation. TTP is an innovative independent consulting company, involved in medical devices, printing, drugs (soft and dispensing), etc. They develop systems, and embedded ones too, and operate on a very clear-sighted basis of what is required, with an understanding of what the markets are looking for and how best they be supplied. The work from the ground up, and see a project through from concept to delivery. Baggage is not carried, but a wealth of expertise is close to hand. So, what is innovative? Customers tend to want their products smaller, cheaper, faster, quieter, with more functionality, wider implementation, and easier user interface; these are the subjects looked at by TTP, and their view is so often from a different perspective, with a mixture of where and when hardware/software can overlap. He gave us some wonderful examples of where and how they had helped their customers overcome competitive technology by a complete redesign and manufacture, within budget. Case studies were described for a digital label printer, a “virtual” sensor, a mobile 'phone, and in all cases it would seem that the client had done rather well, but not surprisingly so.

To compete the afternoon of Day 1, Malcolm Penn looked at the market analysis for chips – hardly a normal market right now. There are four fundamental drivers, all interrelated, but mathematically disconnected, he commented. There are some remarkably strong industry constants. It all fell off the cliff in the last quarter last year, by about 24 per cent, but the second quarter in 2009 recovered, in March it must be said, up 17 per cent. Outlook is for+12 per cent in Q2 and+3 per cent in Q4. The chip industry will recover faster than the economy, said Malcolm, as it was in good shape when it went into it. It was the established economies that were the hardest hit, and it the financial industry that caused it all, not the world trading pattern. Asia is now driving world GDP growth, with a forecast for 2010 of it being up by 2.5 per cent. Financial conditions are better, in places, with bank lending still difficult, in places. There is upside potential, mobile and auto sales held up better than expected. Right now it is impossible to balance supply with demand as there is no pattern, and the old production line manufacturing has all gone to batch manufacture, so forecasting is hard. Thus, units peak in month one, and trough in month three, but the real unit demand is in fact quite constant and 10 per cent long-term IC unit growth means similar long-term growth in capacity. But in Q2 09 capacity was down 14 per cent, and this has to recover.

Front-end Book-to-Bill has never been this low for so long, where capacity investment has been so low, and foundries will be the winners, with IDMs the losers. Such is the downturn that the recovery period will be longer, turning it all off can be done quite quickly, but cranking it all back to speed will take an age. He had 30 years of average selling prices to show us, which was interesting. Malcolm spoke for 60 min, and covered a great deal of ground. He had some strong comments for those who wimped out on manufacture in Europe, he is a passionate European and is very well aware that some bold thinking amongst CEOs is what will maintain the industry citing TMRC as a good example. His outlook for 2010 and beyond was as follows:

• •

  2010:+19 per cent based on continuing recovery momentum.

• •

  2011:+28 per cent based on peak of the structural cyclical boom.

• •

  2012:+18 per cent based on normal cyclical market correction starting 2H-2012

• •

  2013:+3 per cent based on market correction in full flow.

• •

  2014: start of the next cyclical recovery

• •

  Given the impending 2010 fab shortage, the upside for 2010-12 is huge.

Plus ça change (plus c'est la même chose)

Malcolm clearly and volubly showed why he is the industry's leading analyst; there is no aspect of it he does not know intimately, and there is no aspect of it that misses his attention. Here was the definitive view of the past, the present and the future.

Day 2. 2 October 2009

The day got off to a good start with some news from Malcolm Penn about the recovery; the WSTS results had come in overnight and the recovery was on track as forecast. The sun continues to shine, certainly on Geneva, but more notably on the semiconductor industry.

Steve Glaser, VP of Strategic Marketing at Cadence Design Systems, was keen on pursuing the lowering of cost of SoC design. It costs anything between $50-100 million for a SoC design, and these costs are rising exponentially. Losses can be made by delays in coming to market, 51 per cent of costs are at the front end, software is five times the cost of hardware, also there are costly bugs with software and hardware and a combination of the two. Often there is $10 million cost for delays. Management costs are now at about 30 per cent of the overall figure. About 89 per cent of projects are late. It's a minefield.

Steve highlighted various ways in which a safe path could be trod towards successful SoC design coming in on time, and with significant savings, predominantly by the use of a golden source of SoC design, which is a unique platform tuned for high-performance co-verification and optional power profiling at the system level, allied to starting software 3-5 months early, starting design closure three to four months early, and having it all right first time around. Sounds easy, but it probably is not.

Peter Pfluger is the CEO of Tronics Microsystems, whose topic for the morning was MEMS, and he looked at successful business models. MEMS is a technology where carving 3D microstructures into silicon allows microdevices of very diverse functionality to be created, with applications in the mechanical, electromechanical, fluidic and chemical/biochemical fields. MEMS is a market rising in value from $6 billion this year to $12 billion by 2012, and covers no less than 15 different end-use applications. Some of these are accelerometers, micro-pumps, sensor probes, motion transducers, inkjet print heads, motion detector devices; the list is long.

Companies in this field are foundries, component manufacturers, and sensor and sub-system manufacturers. In the top 20 companies lie such names as Texas Instruments at the top end, down to Honeywell, Bosch, Dai Nippon at the lower end. Tronics are doing rather well, with 34 per cent annual growth rate and sustained profitability. The MEMS process landscape includes MEMS-on-thick SOI (20-100μm) for active parts, and stacking here is the key, with sandwiching of several wafers (typically two to four). Trends are towards a generic MEMS process architecture but this is some way off. Volumes of MEMS are low, and costs are therefore still high, but they are mostly specific packages, serving a particular function. Products have a long life, have a high fixed cost, and are never likely to be a high-volume product. But high prices bring with them a high level of global support, which is essential, and the availability of a diverse, essential and highly functional component in the electronics chain seems to be well-worth paying for.

The analog mixed signal foundry of to day and tomorrow is based on expectations of growth in the market, driven by customers, system integration, and by maintaining a competitive advantage, said Michael Lehnert, the CEO of LFoundry. The foundry market is worth $40 billion per annum, and growth is at some 25 per cent per annum right now, although this is expected to drop down to 5 per cent by 2013. Trends are dramatically towards<90 nm. The markets are automotive, as the use of semiconductors in car design increases, and will grow in the years ahead quite steeply. The other market is wireless. Michael went on to describe the differences between an analog foundry model and a digital foundry model, in which the latter appears to be better value, but runs on fixed lines, whereas analog is more flexible but more expensive. However, the system integration of analog, digital and MEMS is needed for system on chip production, and specialised foundry services such as those offered by LFoundry allow increased integration at a silicon level, providing increased complexity, a more efficient design, added functionality, improved power management, reduced packaging costs, reduced test costs, a smaller foot print and reduced I/Os.

Close to the heart of Hossein Yassaie, CEO of Imagination Technologies, is the idea of designing for the upturn. As technology evolves, so too are the consumers, who want everything, now, effortlessly. They have three programmes for creating and licensing intellectual property for their semiconductor partners who create the SoCs and sell the chips to the OEMs who keep the consumers happy. It is a neat circle, but the technology emanates from Imagination Technologies. Hossein sees two trends – one is mobile, “everything, everywhere” and the other is the internet, the ultimate direct link to the consumer. The iPhone is now a “must” in that it has that direct link – the Apple Effect. That link to the consumer relies upon a stable platform, hitherto the PC. That platform is being established in other channels, App Store, App World, etc. Home, mobile and PC is the triangle, and domain knowledge is the USP for anyone. Multimedia will accelerate between now and 2020 at a massively fast rate, through Netbooks, PMPs, 'Phones, DTV/STB and automotive systems. He illustrated the Home Hub, bringing in security and surveillance, domestic power management, assisted living and health, home multimedia. Start-ups in the semiconductor market can succeed if they have sufficient domain knowledge, and related strategy to exploit the available technologies.

Mentor Graphics sent along Joe Sawicki who said that we should not fear the exponential. $120million developments a thing of the past? No one can afford this. Is this true? SoC design costs forecasted to exceed $100 million within three years. Joe insisted that it is the fears that trigger the drive to innovation. As an example, mask design costs in 2009 at 32 nm were less expensive than 45 nm in 2007, through innovation. Innovations in systems design will offset the exponential, through architectural design flow, the flexible exploration of hardware/software architecture, using high level synthesis flow towards verification, and thus a shorter development and verification cycle. Intelligent testbench automation is where input is written in rules, rules are compiled into graphs, and graphs are traversed by algorithms. It is the use of virtual prototyping that will assist design engineers to minimise exponential cost hikes.

Panel discussion

A panel comprising Laurent Boisson, Brian Harrison, Jon Hudson and Joe Sawicki talked about the new corporate challenged in the post crisis world (Figure 1).

Figure 1

Malcolm asked if it was different now, after the crisis, and how should CEOs be thinking. Jon Hudson said that there were some short-term challenges – consumers are going be a lot more wary, and will not blindly follow the latest fads. Blogs reveal the level of discontent amongst consumers, and in the longer term, there is the importance of the SoC, and Asia is where it is and where it will all happen, and we have to focus on getting our technology out there.

Brian Harrison – was it really all that bad? No, things will not fundamentally change. The abruptness of the downturn has caused a shortage of materials and getting restarted may take a little longer, and money will of course, become, eventually, available. But we have to innovate as existing markets become saturated.

Laurent Boisson sees caution in Europe, but a lot of entrepreneurial activity in the USA, and he encourages us Europeans to be more like them. Joe Sawicki added that people are active in the design field in the states, for sure, but the manufacturing side was down turning anyway even before the financial crisis. Think deeply, take risks, that will make you successful.

Is there a green agenda? This has to be addressed aggressively. Is this a new issue? Jon Hudson saw an opportunity here, with innovation on power supplies for sensors, a lot of encouragement for savings in power demand and more efficient and greener sources of power supply. Brian Harrison added that he tends to see power reductions, as a business opportunity also, but when it comes to waste management, there will be people who are not in our industry who will be the innovators in this area.

Laurent Boisson said that improving energy efficiency will be a big field, with solar power, e-cars, etc. It was felt that the industry would do well to drive the green issue forward at every opportunity, it has consumer appeal, and if we waited for legislation from politicians, we might wait a long time for something that would, inherently, be wrong.

Nobuaki Miyakawa contributed to the discussion from the floor. He said that the key words for mega trend would be the environment and productivity to overcome the aging problem in the developed world. The new party in Japan wants a 25 per cent reduction in greenhouse gases which will impact Japanese industry, which Mr Miyakawa sees as a wonderful business opportunity. Changing lighting to LEDs, and electric cars will see a real change in energy usage and demand. Productivity increase is very important, and this is where robotics come into play. Integration density doubles every two years, Honda and Nissan are developing robots energetically, and robotics need a large number of sensors, more than a hundred types of sensors, a big opportunity for SoC devices, also MEMs, and actuators.

Laurent Boisson concluded a free-ranging discussion over various topics with a caveat about capacity. There is a big risk to the semiconductor industry as the capital expenditure required to main the supply side has been too small, and when business recovers there are going to be delays. The future may bright but the traffic might be stuck on orange.

The closing address came from Laurent Malier, the CEO at CEA LETI. Employing 1,500 working on R&D in microelectronics, they are similar in concept to IMEC, but focus more on competitive advantages. LETI supports industry through 3D technology. Looking back over the two days, Laurent said that there has been some good news at this conference, and demand is back in three levels, through incremental revolution, and we have spoken about energy, demands on intelligence, on electronics. We are all looking at new markets and new opportunities, and Europe is not just a service orientated economy, as many profess it to be, but has an industry which is very much alive and well, and the mind set in politics is changing. Laurent is optimistic about Europe, for two main reasons; one is that European industry is really strong in end-user markets, and the other is we have great research institutes, committed to support industry, as well as companies such as IMEC, LETI and The Fraunhofer Institute in Germany. There is much to be cheerful about.

Summary

Given the state of the market, attendance at this unique conference was exceptionally high, but as it is such a dedicated forum, with a loyal following built up over many years, and as it is so well focused upon what the semiconductor industry is, does and wants, non-attendance is really a non-starter. As delegates said, it is always an excellent event, superbly organised, smoothly run, and invariably an enjoyable experience. The when and the where for next year? TBA.

John LingAssociate Editor3 October 2009