Future horizons semiconductor industry forecast seminar 19 January 2012

Citation

Ling, J. (2012), "Future horizons semiconductor industry forecast seminar 19 January 2012", Microelectronics International, Vol. 29 No. 2. https://doi.org/10.1108/mi.2012.21829baa.023

Future horizons semiconductor industry forecast seminar 19 January 2012

Article Type: Exhibitions and conferences From: Microelectronics International, Volume 29, Issue 2

When he is not giving (in)sanity checks to companies, Malcolm Penn, CEO, provides a wonderfully compos mentis insight into the semiconductor industry at one of his unique Industry Forecast Seminars, of which the one on 19 January, held in Kensington, London, was a classic of its kind. Malcolm has interactive and knowledgeable delegates at his sessions, which makes it more than just interesting.

The current world consensus is that “We are all doomed”. No we are not, said Malcolm, as world output in 2012 is forecast to be 4.0 per cent up on 2011, which was 4 per cent up on 2010. The shift which started 10-15 years ago continues, the BRIC countries are bouncing back the hardest, with the US stagnating, rather like Europe, and there is a new economic world order. The emerging countries own 85 per cent of the world population, and the purchasing power of these economies is most important; they are markets in their own rights. Money=power, and China will overtake the USA in GDP next year.

Recessions are triggered by high oil prices, traditionally, but conversely recessions help to lower oil prices, and with the US having 25 per cent of the world’s oil consumption at 19,000,000 barrels a day against China’s 8 m, a change here is inevitable. Yes, there is an inflation problem down the road, with interest rates being as low as they ever have been. The next few years are not going to be very pleasant, what with huge debts, raising inflation, and a stagnant economy.

Malcolm is worried about companies who now treat dealing as a way to make money, in which the time taken to process a trade is more important than actually making anything. The real economic recession has yet to come, and for the advanced economies it is payback time. The IMF is now seeking stronger and deeper ties with Asia to stave off the crisis, but the crisis is imminent.

China and India are looking at a GDP of 9 and 7.5 per cent, respectively, whereas in the Eurozone the GDP forecast shows only 1.1 per cent in 2012.

Rolls-Royce saw 31 per cent growth last year, with the biggest market in the Asia Pacific region. The luxury market in fast-growing economies is traditionally buoyant, but even Bentley were amazed to see over 7,000 of their cars being sold last year in the same area.

The chip market, thought Malcolm, is due to bounce back and rally strongly this year, stimulated by innovation, motivated by a recession. No, we are not doomed, but it is all going to be different. Brazil has replaced the UK as the world’s sixth largest economy, and China will replace the USA as the leading economy by 2013. Yes, it will be very different.

The long term for the semiconductor industry is easier to forecast; when the money ceases to flow, the semiconductor market stops, as economics is all about confidence and trust, and confidence is low and trust has been severely tested. Wafer fab manufacturing will be centred in Asia, as manufacture was always outsourced to Asia, and R&D has followed it. Once that takes place then the hub is there. Not here.

Do not ask Malcolm about Global Foundries, he gets apoplectic. But it led to an interesting discussion about the inability of the industry to manage, and the shortage of the people who know how to. There is a cartel of incompetence sitting on a lid which is holding back a wealth of talent, who needs to be given their head of steam.

Malcolm threw some light on the perennial mystery that is inventory. Why is inventory falling just as demand seasonally increases? There is no contingency, as companies pursue a short-term policy and not a long-term one. Put that against the backdrop of a highly successful International Consumer Electronics Show in Las Vegas last week, then the unreliability of the supply side and its fallibility is another cause of apoplexy. Once inventory is gone, it’s gone. Getting it replaced takes four months, which is not exactly overnight. TSMC makes money out of foundry wafers, but no one else does. Malcolm was concerned that our industry leaders seem unable to understand that average selling prices have been rising in 2011, and base their projections on obsolete and out of date data, and the industry bodies are also singing from the same hymn sheet. It is a disaster if one projects a decline when it is in fact an industry well worth investing in.

DRAMs – not a Scottish tipple, more memorable than that. This is a market which traditionally has never made money (apart from Samsung), and whilst the market is big ($30 b) the losses are significant. So the wiser companies have abandoned this genre and are now making money out of NAND, thanks to the insatiable demand for smart phones. Showing a slide of the market needs, Malcolm explained how it is a matter of markets vs products, Europe is uniquely good at supplying it all, be it MEMs sensors, SiP, integrated passives, embedded RF, advanced CMOS, for markets such as auto, medical, energy, lighting, aerospace and defence, etc.

The bounce back has already started, with TSMC’s Q1 figures, which are excellent, being a harbinger of the industry Q2 figures. Growth rate of 8 per cent in chip sales for 2012, and if this holds true, which Malcolm thinks it will, then we could be looking at 20 per cent growth in 2013, 16 per cent in 2014, but that assumes no natural disasters, and competent political and business leaders. Demand is there. The rest is a worry.

Turning to wafer fab capacity, Malcolm reminded the delegates that there are some basic rules here. It takes a year to get into manufacturing, assuming all goes well, the time or order to delivery is nine months, and the capacity for the next three or four quarters is already spoken for. This is an industry sector that needs strategic vision and commitment, as the peaks and troughs of demand and supply over the years look like a scary ride in an amusement park, except that it is not funny. Wafer fab capacity has changed dramatically since 1991, in those 20 years the Rest of the World has risen from 15 to 65 per cent. Growth in the year ahead will be 6-10 per cent, and whilst IDMs are not investing (apart from Intel, Samsung, TI and Infineon) TSMC thunders on with capital expenditure up 30 per cent, sales up 22 per cent, with Global Foundries and Samsung fighting for the leftovers. What will change the game is 450 nm. Here the winner will take all.

Mike Bryant is the enthusiastic Chief Technical Officer at Future Horizons, and there is little that he does not know about both products and markets. He invited the delegates to consider the key questions for the semiconductor industry – such as the technology that surrounds the ICs we need, who will design them, and who will make them? Processing – Moore’s Law cannot continue, 90 nm was a bad idea, 65 nm did not work at all, and so what hope was there for 32 nm? Well, Intel produced a 32 nm low power Medfield processer which is up and running, and their 22 nm Ivy Bridge processors will be in mass production by April this year. Everyone else is miles behind, Panasonic have yet to show their 20/22 nm node, and of Renesas and other Japanese companies not a whistle so far. TSMC have a 28 nm process in volume production, but UMC and SMIC appear to have given up on advanced nodes.

ASML continues its rise to world dominance in lithography, and is Europe’s most valuable semiconductor company.

ARM continues its impressive list of new releases, with NXP, TI, Freescale, Fujitsu and STMicroelectronics becoming ARM-only suppliers. However, just as ARM become the world standard, so Java and Android have made this irrelevant. Intel is not ARM’s biggest customer, contrary to popular belief, and ARM is busy with new developments which may include network on chip.

Infineon are about to start volume production of power devices on thinned 300 mm wafers in Dresden, and the future looks good for them in the automotive, industrial, chip card and security market sectors; sales for 2010 were $5 b.

STMicroelectronics have a wide range of products and devices, but are handicapped by ST-Ericsson who has French and Italian Government shareholding baggage to contend with. They may have to buy or build a 300 mm fab to maintain its core strength as a broad-line supplier.

In the field of DRAM and Flash, sales of the latter should overtake the former this year. Samsung with 37 per cent market share is shipping 27 nm node with 2-bit multi-level cell product, SanDisk & Toshiba have sent out samples of their new 64Gbit 2-bit device using 19 nm technology. The manufacturing status with Flash is that floating gate technology ceases to work at somewhere between 19 and 15 nm, so with Samsung and SanDisk heading towards 19 nm, a solution needs to be found, likely to be in the short-term, use of Silicon-Oxide-Nitride-Oxide-Silicon (SONOS) gates, and in the longer and safer term it will be SONONOS which adds two additional litho steps. NAND flash capacity growth has exceeded Moore’s Law by using multi-bit cells, either 2 bits or 3 bits per cell. But, as transistor cells reduce there will not be enough electrons in each cell to make this feasible so it will be back to one bit per cell. DRAM sales will continue to fall, and several such fabs are known to be for sale, they could be converted to making MCUs and MtM products.

Litho is the largest manufacturing expense for any IC, and is increasing in complexity as advanced nodes are sought. Even with thousands of DRAM die per wafer, or larger Flash densities, the additional cost is becoming more and more significant. The solution here is EUV litho, which is several years away yet and will quickly focus on 450 nm wafers when it comes. This would allow Samsung to dominate the DRAM market, and once EUV becomes commercially viable then ASML are going to be overwhelmed with orders.

Some of the semiconductor applications for this year and onwards are interesting. Apple will launch the iPhone 5, but Nokia will have to do better than their Windows phone; Patent issues will distract companies from making better products, but Ultrabooks will find their place in the market; a large number of business machines are due for replacement this year but many will wait for the Windows 8 release in October. IPad3 will launch, with or without a 3D display, and Kindle Fire will become the No. 2 machine once launched.

SmartTV and 3D are keeping the television market healthy and growing, with increasing internet delivery, and SuperHD (8 m pixels) and UltraHD (33 m pixels are now undergoing trials. Apple may also be moving into the TV market with a box that can turn a dumb monitor into a more advanced SmartTV than anything else on the market. It could also run all iPhone/iPad apps, possibly using some form of hand-held stylus. It could also be an alternative route to buying a Mac.

Amongst the pictures emerging from Mike’s crystal ball, in which he looked forward to as far ahead as 2025, were implantable mobile phones, implanted health monitoring systems, 3D projection displays, using graphene-based emitters; quasi-intelligent robotics; self-powered sensors with short range wireless which can monitor anything and everything.

By 2013, a little closer to the present day, Mike was optimistic that HMG would auction off 4G bands, three years after the rest of the world. Of the six European flagship projects, two would begin operation, each with a budget of around €100 M per annum, and energy harvesting devices will become mainstream inside many products. He had a rather detailed list and it will be fascinating to re-visit them in 2025 and see what did and did not take place. But it will not be this writer!

Malcolm Penn summed up the day with the view that 450 mm will divide and rule, that companies have to stop and think about out-sourcing production, and off-shoring jobs, there is real cost here, and that chasing the fast buck never built anything sustainable.

If European politicians begin the long process of deciding if and how Europe can remain in advanced semiconductor manufacturing, then there is hope. If any of the ENIAC projects can be brought successfully to commercial completion, then there is hope. Europe may not manufacture en masse, but the numbers of potential customers for innovative products do give hope that Europe can be a player in the global market place. For these, semiconductors will be vital. For those involved, such Future Horizons seminars such as this are equally as important.

Future Horizons next Industry Forecast Seminar will be held on 17 July in London, and their 21th International Electronics Forum will be held on 3-5 October at the Marriott Hotel in Yerevan, Armenia.

John LingAssociate Editor