APPENDIX 6: VISIT TO SILICON VALLEY
1. On 10-12 June 2002, the Sub-Committee visited
Silicon Valley, the concentration of computer-related business
some 40 miles to the south of San Francisco, California which
has helped give the US its leading role in microprocessing matters.
The purpose of the visit was to enable members to discuss with
a range of US experts not only the exploitation of CMOS technology
to its limits but also possible post-CMOS technologies
all with a view to better equipping the Sub-Committee to consider
further UK contributions to the global computing industry.
2. The visiting party consisted of Lord Wade
of Chorlton (Chairman of the Sub-Committee), Lord Methuen, Lord
Mitchell, Lord Oxburgh and Lord Patel. The party was supported
by the Sub-Committee's Specialist Adviser (Professor Steve Furber)
and Clerk (Mr Roger Morgan), and by members of the British Consulate-General
in San Francisco.
Stanford Seminar (MONDAY 10 JUNE, MORNING)
3. The visit began with an academic seminar arranged
with the assistance of and at Stanford University. (This was essentially
a US counterpart of the seminar held at the Royal Society on 13
March[114].)
a. James Plummer, Dean of Engineering at
Stanford University, welcomed the Sub-Committee and other participants
to the University. In setting the scene for the seminar, he noted
the technical challenges set out by the ITRS if CMOS technology
was to be exploited to its theoretical minimum gate-length of
20 nm over the next ten to fifteen years. He saw the challenges
as essentially manufacturing questions which were intrinsically
soluble, although there remained a question about the costs involved
in implementing the solutions. There would be potential for non-CMOS
devices, but he felt the dominance of and investment in silicon
technology meant that new devices would be successful only if
they offered significant new markets. None would be a straightforward
replacement for silicon which had defined and constrained our
thinking about systems. In the meantime, enormous opportunities
seemed to be presented by better design and utilisation of CMOS
technology.
b. Dr Larry Dalton of the University of Washington
described his work on organic electro-optic technology. This was
entirely compatible with CMOS and enabled direct Internet and
other communication access at enormous speeds of up to 100 GHz.
The work had been funded by NSF grants and had led to great interest
within the US and elsewhere. A framework for exploiting the IP
was proving more complicated than anticipated. Work was still
in progress on that.
c. Dr Meyya Meyyapan, Director of the NASA Ames
Centre for Nanotechnology, outlined NASA's grand challenge of
using revolutions in electronics and computing to create small
(less than 10 Kg) autonomous or thinking spacecraft for exploration
within the Solar System. Effectively, this required the development
of teraflop computing within the dimensions and power consumption
of present laptops. NASA was pursuing a number of nanotechnology
initiatives which might help solve the formidable problems in
meeting that challenge.
d. Dr Evelyn Hu of the University of California
at Santa Barbara noted that the death of CMOS had been predicted
for many years. It was, however, clear that the physical limits
of CMOS technology were now in prospect. She agreed with Dean
Plummer that the development of CMOS might ultimately be constrained
by economics rather than manufacturing capability. Spintronics,
molecular electronics, quantum computing and self-assembly all
seemed to hold some prospect for leapfrogging over the CMOS limitations.
Although success was far from guaranteed, their time might come
and it was essential to maintain a portfolio of research balanced
between exploiting CMOS to the maximum and potential new approaches.
Work on the latter might, in any case, usefully feed back into
the further development of CMOS with which, in any case,
it seemed likely that any new technologies would need to be integrated
to at least some level.
e. Dr Erich Strohmaier of the Future Technologies
Group at the Lawrence Berkeley National Laboratory surveyed the
field of supercomputing. Advances had been phenomenal: current
laptop computers would have made the list of the top 500 supercomputers
only eight years ago but annual performance in supercomputers
had been growing by a factor of about 1.8 a year. Further progress
was possible through technological and design improvements and
was needed to meet developing needs for large-scale modelling.
One of the constraints on supercomputer development (and on computers
more generally) was the interconnect problem. Super-fast processors
could not be used to their maximum unless data transfer in and
out was radically improved.
f. Dr Sung-Mo Kang, Dean of the Baskin School
of Engineering at the University of California at Santa Cruz described
the University's Centre for IT Research in the Interest of Society
(CITRIS) established in partnership with Industry and Government
to explore matters of mutual interest. Noting the potential for
wide-ranging applications offered by further refinements in CMOS
technology, he outlined the need to match further miniaturisation
with appropriate low-power techniques.
g. Professor James Demmel of Berkeley University
described the grand challenge accepted by CITRIS to develop low-power
miniaturised sensors with built-in wireless networking ("smart
dust") which could be used for a variety of purposes. The
data gathered from these devices (and, to avoid data overload,
filtered by the collecting networks) could be used for a variety
of social policy purposes, for example, in relation to energy
efficiency, environmental matters and health care. CITRIS was
confident that the technical challenges could be met by using
state of the art CMOS technology, and had already produced some
prototypes. There remained some issues about the political framework
within which such wide-spread monitoring would be acceptable and
in relation to handling the IP arising from the project.
4. In a brief concluding discussion, the following
main points were noted.
a. There was a great deal of life left in
CMOS technology to be exploited both through technical advances
and better design and architecture.
b. Nevertheless, the physical and economical
limits of CMOS were in prospect. Proportionate resources should
be applied to blue-skies research on successor devices.
c. Generally, business had too short-term a view
of research and development. Public funds would continue to be
required to underpin the essential basic research.
d. Commercial impact came about not from technical
advances themselves, but from the products they facilitated. Developments
were often better driven by grand challenges than specific technical
objectives. In this regard, biotechnology seemed a promising field.
Sandia National Laboratory (MONDAY 10 JUNE, AFTERNOON)
5. Dr Mim John, Vice President of Sandia National
Laboratory, and her colleagues outlined the work undertaken at
the Sandia and other National Laboratories to develop extreme
ultra-violet lithography (EUVL). The National Laboratories concerned
were originally established for defence purposes and this was
still their main activity. In recent years, however, they had
diversified and the EUVL project was one of a number of non-defence
activities.
6. There had been awareness since the early 1990s
that the present generation of lithography technology would, at
some point, fail to provide the resolution required for the fabrication
of chips with ever smaller components. EUV, in which the Laboratories
had some expertise from earlier basic research was one of four
possibilities contemplated by semiconductor interests, the others
being X-rays, ion beams and electron beams. EUVL had become the
front runner.
7. EUV was easily absorbed. EUVL had therefore
to be conducted in a near-vacuum and with appropriately-surfaced
reflective mirrors in place of the transmissive mask and lenses
of present machines. A novel business model had been developed
with leading microprocessor companies (including Intel and IBM)
to develop the technology. This ensured not only that the risks
were spread but also that there would then be a sufficient market
for the resulting technology. A prototype machine had been completed
and had met all the specifications. EUVL offered such advances
in lithography that it would be "good to the end of silicon".
Intel (TUESDAY 11 JUNE, MORNING)
8. Dr David Tennenhouse, Intel's Vice President
for the Corporate Technology Group and Director of Research, and
his colleagues traced the development of computing from the early
scarcity of machines to today's position where one person had
to interact with many computers and computerised devices. They
saw the future as truly ubiquitous and human-centred computing
in which those many computer devices would be networked and interfaced
so that they served rather than overwhelmed the user.
9. The technical challenges of ubiquitous computing
could, in Intel's view, be met entirely by further development
of CMOS technology although there remained a number of
substantial challenges, not least in managing power consumption/heat
dissipation. Design considerations would become even more important,
and it was a matter for concern that universities' computer design
research had declined in the US and elsewhere.
10. The key obstacle at present was the non-technical
matter of enabling users to treat computers as a trusted client.
The need was not just to develop the technology, but the products
which it facilitated. The key driver would be enabling the user's
intuitive and natural interaction with personal and embedded computers,
leading to truly personal computing.
IBM (TUESDAY 11 JUNE, AFTERNOON)
11. Dr Robert Morris and Dr Robin Williams[115],
respectively Director and Associate Director of IBM's Almaden
Research Centre, and their colleagues together with Dr David Cohn,
Director of IBM's Austin Research Laboratory, surveyed IBM's wide-ranging
research programme. They noted, as others had done, that the limits
of CMOS technology might be dictated by economic rather than technical
barriers and that, as CMOS reached its limits, there were substantial
challenges to be overcome, particularly in power management. Nevertheless,
CMOS would remain a formidable technology facilitating the development
of many new and improved devices.
12. IBM's research was addressing not only the
further exploitation of CMOS and related technologies of memory
and interconnect, but also non-CMOS technologies, including nanomagnetics
and nanotechnology[116].
They had achieved 7-qubit quantum computing using specially designed
molecules in which the quantum states were manipulated and measured
using magnetic resonance techniques. Other technologies would
be needed to achieve useful numbers of qubits. With associated
advances in programming, quantum computing could quite simply
rewrite the rules for computing, particularly in the field of
encryption.
13. Generally, IBM saw the possibility of market
saturation for client computers, but scope for an explosion of
embedded computing. CMOS and associated technologies had largely
become commoditised. Value remained to be extracted from the areas
of design, architecture, software and services.
Finance (WEDNESDAY 12 JUNE)
14. Richard Irving[117],
Founder Partner of Pond Venture Partners Ltd; Michael Wishart,
Managing Director of the Investment Banking Division of the High
Technology Division of Goldman Sachs; and Nick McKeown, Associate
Professor of Electrical Engineering and Computer Science at Stanford
University (and co-founder of the successful internet-related
Abrizio Inc) outlined the financial and legal structures within
Silicon Valley that facilitated the lively technology transfer
from academia and spin-out from other companies.
15. Combined with general culture (in which failure
was widely seen as a positive and useful experience), the village-like
atmosphere meant that there was a good understanding in all quarters
of others' needs and perspectives. This was reinforced by wider
public dialogue about developments in technology. While many venture
capital firms in the United Kingdom and elsewhere had excellent
financial judgement, Silicon Valley firms were characterised by
a good understanding of the technological issues as well. This
enabled entrepreneurs to extract maximum value from the infrastructure
that such firms provided.
16. The computing industry was global, but had
a substantial nexus in Silicon Valley. It was not self-promotion
to make the point that UK and other non-US companies wishing to
break into the global market would benefit from a presence in
Silicon Valley. In return, Silicon Valley provided not only a
range of expertise, but also CEOs and others with invaluable experience
of developing high-tech products.
Thanks
17. At each event, the Sub-Committee thanked
the hosts for their hospitality and others involved for the opportunities
to explore the various issues. All those visited said that they
too had found great value in the exchanges.
18. On departure, the Sub-Committee warmly endorsed
the Chairman's thanks to Roger Thomas, British Consul General
for San Franciso and the surrounding area, and his staff
particularly Dr Sharima Rasanayagam and Dr Henry Mahncke (respectively
Consul and Vice-Consul for Science and Technology) for
their help in setting up the visit and their excellent support
throughout the programme.
114 See Appendix 5. Back
115
From whom the Sub-Committee had heard oral evidence at its hearing
on 22 May 2002. Back
116
On the day of our visit, IBM's Zurich Laboratory announced the
prototype Millipede, a nanotechnology device (using punched card
principles) that offers storage densities about 25 times greater
than current hard disks. Back
117
Pond Venture Partners also submitted written evidence to the Inquiry,
see page 202 of Volume II. Back
|