Further evidence of the European Commission's interest in re-invigorating microelectronics on the European continent and microelectronics manufacturing in particular has come from Neelie Kroes, European Commissioner responsible for the digital agenda.
Kroes, one of the senior bureaucrats within the Commission, used the opportunity of a keynote speech presented to the IMEC Technology Forum, held in Brussels last month, to ask whether Europe should consider the creation of an Airbus of chips.
Readers may remember that Malcolm Penn of Future Horizons Ltd. was one of the co-authors of a report recently presented to the European Commission that discussed various requirements for, and ways towards, hosting manufacturing on 450-mm diameter wafers. However, while interest is high in Brussels the home of the European Commission because of its potential job- and wealth-creation benefits, the idea seems to be less urgent for a number of European chip companies that are, of course, driven by nearer-term financial considerations.
Meanwhile, data from World Semiconductor Trade Statistics and elsewhere has shown that the European manufacture of chips, and market for chips, have both been in decline for a number of years.
Kroes raised the level of rhetoric by asking whether Europe wants to be a global player or not, and whether it would not be better to opt for European consolidation and cooperation "on our own terms" before consolidation is forced upon European companies for which one might read: "forces them out of existence."
In the past the European discussion has mainly been about collaborative R&D, billions of euros of which is already supported by the Commission. Now Kroes has cited the more interventionist example of Airbus SAS, one of the great European projects and one that has achieved considerable success.
Airbus is a subsidiary of European Aeronautic Defence and Space Company NV (EADS), which was deliberately assembled from national defense and aerospace companies across Europe to provide strategic defensive independence at a continental level. It was also set up to produce a rival to Boeing at a commercial level, which it has done with some success. Airbus now produces about half the world's jet airliners.
Does anybody else remember European Silicon Structures? Any lessons to be learned from that experience? I think it can hardly be claimed as a success, but maybe that was a case of too many innovations at once rather than too many Euro-cooks spoiling the broth.
I fear that the big, lumbering bureaucracy of Airbus, riddled with political compromise, fits in perfectly with the aerospace industry but would not last long in semiconductors
- Risk-averse culture (because of big government?)
- Weak funding mechanisms (much smaller Seed/Bridge/Business Angel/VC pool of money compared to the US)
- Fragmented European markets e.g. when a bright kid in the US thinks of a social networking business, he/she has a 300-million market to start with, a European kid starts with 80 millions at best)
Here is an interesting parallel. In UK after WWII there were four private regional railways all struggling to make any money after years of neglect and unable to face the cost of upgrading to diesel/electric operation.
On the first sniff of possible government compensation the companies were eager to take the money and be nationalized out of existence.
Before they do so they better consider the arrogance of the systems engineers at Airbus. their attitude is that they know how to handle emergency situations better than the pilot, therefore when an unusual situation comes up the autopilot kicks in regardless of whether the flight crew wants it to or not.
This is was brought Air France 447 in 2009. The pilot could not control the aircraft because the autopilot kept kicking in and taking control based on faulty readings from the pitot tubes (airspeed indicators), which had iced up. Whenever the cockpit crew turned the auto pilot off and tried to correct the aircraft's flight, the flight computer assumed that they were incompetent and therefore re-asserted control of the aircraft from them. With the pitot tubes iced up, the computers thought the plane was flying too slow so it put the aircraft into a dive to pick up speed. The pilot would turn the autopilot off, pull the plane back up, only to have the damned thing re-assert itself and put the plane into a dive again. Without a master override, he couldn't keep the plane from destroying itself. Eventually it crashed into the ocean. This is also the reason the Airbus 380 which lost an engine to a fire almost crashed; the autopilot assumed the situation was different and kept taking control from the pilot when it felt the pilot was acting foolishly.
Boeing aircraft don't do this.
"We have the academic and innovative capacity to take part, but we seem to lack the will to invest in the manufacturing part" - this is called the European paradox, a phenomenon that eurocrats have tried to believe for ten years but that has been identified as largely a myth in more recent studies (http://en.wikipedia.org/wiki/European_paradox).
The facts are that the combined public and private R&D investments in Europe are much lower than in the US, that the market is still very fragmented in national countries and taking risks is not rewarded.
A key difference with the Airbus case is that the aerospace sector was and is much more dependent on the government and so much more amenable to large-scale 'forced' consolidation by these governments - which was probably a very good idea, as Airbus shows.
Ms. Kroes usually knows what she's talking about, and a European chip technology company might make sense, but a central problem remains: most companies creating fundamentally new markets over the past decades are not European ones (Apple, Google, Microsoft,...) and that's not likely to change any time soon.
An airbus of chips? Late for delivery, plagued by subcontractor problems, not always tested adequately, and delayed by parts shortages, in addition to being single sourced? I suppose that if the larger substrate could be made much larger, that there may possibly be some benefit if the yield wound up being adequate. But the cost of a bad substrate would be MUCH larger, it appears.
"But if the likes of ST and Infineon, with their public shareholders, don't want to be conglomerated, what's to be done? "
That is why I said Eurocrats should "facilitate" not necessarily drive European chip integration. The lever to do this is a mixture of carrots and sticks e.g. direct financial incentives for integrated efforts e.g. in new technology nodes, and financial disincentives for direct competition in certain segments e.g. DRAM.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for todays commercial processor giants such as Intel, ARM and Imagination Technologies.