Future Horizons Mid-term Semiconductor Industry Forecast Seminar Kensington, London 12 July 2012

Future Horizons Mid-term Semiconductor Industry Forecast Seminar Kensington, London 12 July 2012

Article Type: Exhibitions and conferences From: Circuit World, Volume 38, Issue 4

Malcolm Penn, CEO of Future Horizons, welcomed everyone to the seminar, held in London.

The global economy. Ah yes, it is a mess. Nothing has been sorted out at all; the summit in Mexico was notable for there being no agreement at all, about anything apart from having another meeting later on. The US National Debt is increasing by $4 billion per day, just to put things in perspective. Settlement day was not mentioned, but the bill stands at $15,885,288,877,247.46. Europe descends into a morass of mutual indebtedness, with a rapidly shortening lifespan.

World GDP growth is forecast at 3.9 per cent, which is above the long-term average for the last 20 years. And 4.1 per cent next year, again well up on the long-term average. But where is the confidence that goes with it? More seriously, those responsible for the financial woes in which we presently find ourselves are still sitting at the top table. The Head of the IMF Christine Lagarde is no visionary, it would appear, she exhorted Member States to work together, as an original idea it rates 0/10. To make matters worse, the Euro is overvalued.

On a more down to earth basis, Malcolm suggested that we need to plan ahead; inactivity is not an option. The current outlook is rosy, and the world economic forecast looking at 4.1 per cent growth next year. Remember that 2011 was a year of disasters, but the world GDP still grew at 3.9 per cent.

Industry is sitting on cash, but investment is not taking place, new industries are not starting up, and the European chip market is 15 per cent down. It might have to get worse before it gets any better. Swings of swings of 60 per cent are normal in this industry, and as we have been living beyond our means for so long, the repayment of credit will be a painful and slow process. “Neither a borrower nor a lender be” wrote William Shakespeare in “Hamlet”.

Better to have capacity early, but that needs long-term vision. The period of living off inventory is now over, and sales are coming from demand for capacity. All of which takes time, and the amount of time it takes does take does not vary one iota. It takes a year. You cannot hurry capacity, from build to full production it takes 29-36 months.

The long term trend on selling prices is flat. They were going up until the Euro crisis came along and then they fell back. So what is the 2012 number? In January Malcolm said 8 per cent increase in sales, but it turned out to be 4 per cent quarter to quarter, reflecting the delay in the market rebound, and we go into the 3Q with a negative attitude, and no inventory. In 2013, the growth will be 15.6 per cent, 2014 16.0 per cent and 2015 3 per cent. We will see.

Wafer Fab capacity – capacity is fixed for the next three months based on decisions taken nine months ago, and the way in which capacity always fails to match demand is one of the mysteries of this industry. Monthly shipments vary dramatically, so in order to buy smoothly the buyer needs to be able to tell the supplier what he needs 12 months ahead. But who does that? The dynamics are now worse than before by a factor of three. No one knows what they want and when, and they will not commit. If they did it might be costly but it would be so much more efficient. ASML have been listening to Malcolm, so their supply situation is nowadays much better than before, and which proves that it can be done. 450 mm might be easier, as less people are involved, thought one delegate.

Mike Bryant, the enormously knowledgeable CTO of Future Horizons spoke for the afternoon session – and regaled us with the story of how they had to work with a French partner on an EU project entitled SMART 2010/62 all about 450 nm semiconductor, which was completed within one year (2011). Intel proved the point by moving into 450 mm processing just before they began work, which was a sure sign that they were on the right track. The cost of a volume 450 mm fabrication plant will exceed $10 billion! The demand for output from advanced fabs will increase, so by 2018 this will be a reality. Prototypes will be around in 2015, and production ramp-up will be optimised by 2016. Intel is expected to build three 450 mm fabs in Oregon, Arizona and Ireland. TSMC will upgrade their Fab 12 to 450 mm as soon as possible and Samsung will also want to be involved somehow and somewhere. Global Foundries will follow soon after and by 2020 these four companies will represent 70 per cent of current semiconductor capital expenditure, rising to 80 per cent by 2030. Fabs are not major employers; most of the manufacture is fully automatic, with just a few highly skilled technicians. But the impact upon a country’s economy is profound.

Europe has lost out on this type of installation; there was an anti-IMEC feeling in Europe, from a few known companies, who should have known better. However, Europe will still be a key contributor to 450 mm technology, and €100 million will be invested by the Dutch and Belgian Governments in IMEC and ASML; ASML being the key lithography equipment provider, and IMEC the key provider of R&D for processing, materials and transistor structures.

The use of 450 mm wafers will lead to a 28 per cent saving in die cost, due to the reduced number of processing steps. 8 nm will be the first node that is exclusively on 450 mm wafers, instead of the proposed 300 mm wafers on 450 mm equipment. 8 nm is seen as the post-CMOS era. Global Foundries will be purchasing IBM in due course. The 450 mm rollout will not be as fast as some would like, but too fast for others! Qualcomm and nVidia are building business plans based on the savings from 450 mm processing. IMEC is the best place in Europe for semiconductor supplies; it is Europe’s sole hope for success.

Mike went on to talk about other companies, such as Nokia, which is a tragedy. A self-inflicted tragedy as well. Once the company that showed Europe at its best, with excellent product design combined with innovative ideas. But it was overtaken by others with smarter technology. Can it be saved? It seems doubtful. ST Ericsson is another company that has suffered grievously. Intrinsically linked to Nokia has been the (mis)fortunes of STE, a company making products exactly to their core customer’s requirements but this leading to poor sales elsewhere. In turn this led to a lack of economies of scale so the core customer looked elsewhere for dual sourcing, so that ST-Ericsson were then forced to match those prices, but to do so had to cut R&D on new products. Many parallels to GEC-Plessey Telecommunications in the 1990s.

Mike listed some other salient matters, such as the fuss over PV and new battery technologies dying down, and that the UK Government, whilst actively supporting nanomaterials fails to grasp the significance of graphene, which is a sure-fire winner and requires probably no more than a 100 million to get it up and running. There are a few bright spots still offering some support, such as Scottish Enterprise, the South Yorkshire Fund, NESTA, and the EV Group. Throughout Europe things appear even worse, and what is irritating is the development agencies that are still wedded to the 1990s model of attracting inward investment yet cannot even raise the money for a “plane trip”! Alarmingly, people are no longer a sought-after resource, said Mike. Both the USA and Israel appear to be the only places significant Venture Capital (VC) activity is happening. If you want to do a large-scale start-up it is time to pack your bags. The economy does not stop scientists and engineers having bright ideas, but with the general lack of funding most ideas are never going to reach commercialisation. However, this does not make the idea in itself worthless. Au contraire, one route, perhaps not for everyone, is patenting the idea and then licensing it to a third party. Originally the British Technology Group would have been a low cost option, especially for universities, but this now specialises solely in healthcare, thus the best patent licensing companies are now invariably US-based. For a UK based university professor or independent engineer with a great idea but no money to either commercialise it or even to attempt to licence it to a large company themselves, this is often the only route to making some money from it, and you will still have to sell the idea to the licensing company and to get the best terms this will need to be as detailed as any VC Due Diligence report.

Malcolm Penn brought the day to a close with his view of a changing industry, which will look a very different place when we look back in ten years’ time. The EU Commission is facing a divided Europe, divided between 450 mm against the others, but only 450 mm will allow Europe’s indigenous chip firms to catch up their lost leadership position in advanced CMOS manufacturing, placing them at the forefront of technology both in MM & MtM. Embracing 450 mm now will ensure a clear migration path for all future silicon-based chip processing into the foreseeable future. If Europe’s indigenous chip firms ignore the 450 mm paradigm shift, focusing instead on just MtM not MM, they will end up in a technology dead end, their MtM expertise slowly cannibalised by more advanced technology-based firms looking to re-use their depreciated (n−1) MM platforms and shrinking remaining markets squeezed by ever-increasing over-crowding amongst their similar technology peers. Embracing MtM without MM will undermine Europe’s long-term KET aspirations and advanced manufacturing needs by 2025.

The next decade should be brighter. We do hope so.

John LingAssociate Editor