Posts Tagged ‘ IBM ’

X-bit labs – Symbiosis Between Globalfoundries and IBM Needed for Long-Term Success – Analyst

X-bit labs – Print version.

Close Collaboration with IBM Can Provide Globalfoundries Advantage in 20nm and Thinner Nodes

by Anton Shilov
05/11/2010 | 11:05 PM

The emergence of Globalfoundries could be one of the most significant events in recent semiconductor history, an analyst said. While for Advanced Micro Devices the creation of Globalfoundries resulted in cost savings only, for Advanced Technology Investment Company it is the chance to create the world’s most powerful contract maker of chips. However, in order to achieve this, Globalfoundries will need to work closely with IBM and adopt its chip design tools.

Last week it was reported that Globalfoundries could acquire semiconductor manufacturing operations from IBM, which would provide it additional clients, manufacturing capacities and, perhaps, certain intellectual property. But while it is crucial for Globalfoundries to boost its advanced manufacturing capacities so to be able to compete against Taiwan Semiconductor Manufacturing Company already in the mid-term future, for long-term success it is very important for Globalfoundries to collaborate with IBM in general and adopt/deploy IBM chip design tool-systems, which would give Globalfoundries a decisive competitive advantage in 20nm, 14nm, and finer nodes, according to Boris Petrov, managing partner of the Petrov Group.

“IBM will maintain its integrated circuit (IC) process technology leadership via research, but the critical business requirement is also that its Common Platform silicon alliance continues to be successful. […] To be successful Globalfoundries would have to meet cost economics that IBM has apparently failed to meet. This evolution stage represents an immense opportunity – if Globalfoundries, jointly with IBM, is able to construct and implement a new and differentiated vision,” Boris Petrov has written in a new column dedicated to Globalfoundries.

IBM Design Approach: When Perfection Means Isolation

The three primary areas of concern to an electronic system designer are power, timing, and noise. An optimal design technology addresses them in an integrated manner; such a system approach is the spirit as well as a distinctive differentiation of IBM’s chip design approach.

“The foundation of IBM’s leadership position in technology-based services is IBM’s focus on automation; in the case of ICs it is IBM’s focus on automation of system-level design processes. Before actual implementation in silicon, IC design entirely resides in software – at the system architecture, modeling, and application levels. Such software-based IC designs and their design tools are among the most complex software ever developed, and their complexity will continue to increase,” said Mr. Petrov.

IBM’s “abstraction engines” model basic concepts (shapes, timing, other) at such high levels that they are also used in IC-unrelated modeling (financial, materials, biological, other), notes Mr. Petrov. As chip designs become bigger and more complex, such an approach will be more and more compulsory for successful “first-pass” design with billions of transistors in 28nm, 20nm, and finer lithography technology nodes. The recent woes with TSMC’s 40nm and potential issues with 32nm have already cost chip designers millions of dollars, forced TSMC to can its 32nm fabrication process and the virtually the whole industry to reconsider the roadmaps. But nothing is likely to limit demands for higher-performance computing and going forward fabless designers of chips will have to work closely with foundries and the latter will have to concentrate on creation of design tools, which ensure that advanced designs can be made in high volumes and on time.

“The chip design factory approach to silicon integration will likely be the cornerstone of the sub-40nm semiconductor industry. In the sub-32nm chip designs, the emphasis decisively shifts away from an individual expertise and tools approach (the “presence of a super-engineer” concept) to a tightly integrated chip design factory approach,” explained the analyst.

IBM’s IC design focus continues to be on the needs of state-of-the-art technology, still the center of the chip business has moved away from proprietary modeling and toward open systems which are mandatory for adopting third-party intellectual property and creation of third-party chips. Verification flow, making designs manufacturable without having to model down at the transistor level, and power and timing closure in 28nm and finer lithography all present immense new challenges, the analyst stresses. IBM has already expanded and integrated its tool systems with industry standard tools for commodity solutions. Nonetheless, the overall concept remained unchanged: IBM’s tool systems continue to be aimed at the leading edge chips and third-party partners maintain and support the older tools.

What is important here is that only a handful of companies – including, but not limited to, AMD or IBM itself – require state-of-the-art fabrication process or designs. As a result, for IBM, its focus on perfection means isolation from the volume market. As a consequence, despite its advantage in design systems, IBM has had limited success outside internal use.

From Extreme to Mainstream

The mainstream merchant market’s cost and IBM’s profitability margin requirements are too far apart, therefore, it is unlikely that IBM will put much more efforts into development of its foundry business. IBM’s cost structure and focus on its own demands often make IBM the IC design partner of last choice: a client selects and pays for IBM services because it has nowhere else to turn and since IBM provides an expensive guarantee of on-time delivery of differentiated chips.

On the other hand, the chips that contain billions of transistors and considered “extreme” today will become mainstream tomorrow and companies developing them will have to use chip design tools that not only support such complexity, but ensure their low power consumption and introduction on time. Complex devices – such as central processing units or graphics processing units – tend to increase their transistor counts rather rapidly and in less than ten years time there will be chips containing tens of billions of transistors. Needless to say that Globalfoundries and other contract manufacturers will have to provide tools to develop chips of that complexity and potential acquisition, adoption, and deployment of IBM’s chip design expertise and suite of IC design tool-systems will be just what the doctor ordered for the company.

“The time for full demonstration of the power and superiority of IBM’s [chip design] approach is perhaps ahead. Perhaps, it will be the only approach possible in advanced lithography, with ICs with tens of billions transistors,” said Mr. Petrov.

In case the analyst is correct, then, if IBM sells its tools to Globalfoundries, the latter may find itself in a much more competitive position in years. Perhaps, with IBM’s suite of chip design tool-systems Globalfoundries may become the only contract maker of semiconductors, who can produce state-of-the-art chips with tens of billions of transistors or at least it will be much more ahead of its rivals.

Globalfoundries Should Convert IBM’s Design Tools for Volume Production

“To successfully deploy IBM’s IC design tool systems and expertise to much larger and rapidly growing segments of the consumer market, Globalfoundries would have to be able take the good and differentiated and to reject the obsolete and gold-plated,” said Boris Petrov.

At present Globalfoundries is fighting for manufacturing volumes via expansions of capacities as well high yields of chips made using leading-edge process technologies. But going forward – as chip designs get even more complex whereas mainstream customers will be unable to design them from scratch – Globalfoundries will have to provide complex support along with robust services, which is when/where IBM’s technologies of today will be required. The difficult challenge will be to drop the too expensive technologies and convert immensely valuable technology into fiscal gold.

“A key implication of Globalfoundries and the industry’s evolution is that chip design is becoming synonymous with an industrial robotic factory. System vendors need tightly integrated chip design and wafer foundry factories. If Globalfoundries is able to obtain, adapt, and cost effectively deploy IBM’s chip design capabilities it will have a decisive and sustainable competitive advantage in advanced technology nodes for its foundry customers,” asserts the analyst.

Advertisements

Globalfoundries impact and evolution could be significant, says Petrov Group – Part I

Globalfoundries impact and evolution could be significant, says Petrov Group – Part I.

The emergence of Globalfoundries (GF) could be one of the most significant events in recent semiconductor history. While the new company faces significant near-term operational and, especially, organizational challenges, there are also several likely strategic actions that GF will undertake to enable its evolution and transformation, according to Boris Petrov, managing partner of The Petrov Group.

Among several major stages in GF’s evolution, two are quite predictable. The first stage could be near term: the acquisition of IBM’s IC fabrication facilities. There is another and related stage of GF’s evolution that is possibly more significant but also more difficult to implement: GF’s potential acquisition and adoption of IBM’s IC design expertise.

Evolution Stage One: IBM and Its IC Fabrication Business

Why would GF take on IBM’s semiconductor manufacturing, which struggles to turn a profit even in good years? Why would this be a good fit for GF? Is this what GF needs most among its operational and strategy challenges that are threatening its very launch? What will it accomplish by taking over IBM’s semiconductor manufacturing? To clarify these questions one should first analyze the changing role of the IC fabrication business at IBM.

IBM’s core businesses are systems and engineering services; IBM is a technology-based solutions company that functions at multiple economic and geopolitical levels worldwide. IBM’s primary challenge is how to profitably double in size and grow to a US$200 billion company within a decade or so. IC fabrication is unlikely to play any role in the solution to this complex corporate challenge.

IBM’s “DNA” and culture strive to automate business processes of any kind; this is a core competency of IBM that allows IBM to embrace and penetrate all major businesses. IBM’s research has been a fountain of basic invention, often of entire businesses and industries; its US$3 billion Research division is a growth foundation for the entire company. It has remained unsurpassed by any commercial or government research institution; it is the benchmark that defies the common belief that creativity and innovation cannot excel within a large corporate bureaucracy.

IBM is also a military-like business machine, always at war and despite what other companies may publicly say, IBM is typically 5-7 years ahead of the industry in strategy formulation. For all its positives, IBM also has considerable challenges that plague the company. For example, it has lost market share in the IT business for many years.

IBM’s semiconductor business, including IC fabrication, was historically a key strategic element to its entire server system and software businesses. In the semiconductor industry, system companies have been IBM’s customer engagement targets because IBM can enable them technologically along the entire 360-degree silicon integration continuum (from concept design, to silicon, to board, to end-product). Here IBM stands alone with an array of advanced technologies in each segment of the silicon integration continuum-a one-stop technological weaponry-shopping place.

However, there has been a shift in the electronics industry – from Computing to the Consumer IC sector, where low cost is the primary requirement. If we broadly define the Consumer sector and include segments of cell phones, notebooks, netbooks and games consoles, such a Consumer-like sector will soon account for 60% of the total IC industry. But, low-cost IC fabrication is not among IBM’s core strengths.

The Petrov Group projects that IBM’s US$3 billion Research division will continue to drive IBM’s evolution as well as the evolution of the entire IC industry – and much beyond silicon. IBM’s material science and microelectronics research will not only be maintained but also accelerate. However, to accomplish its research and corporate growth goals IBM no longer needs IC revenues that have been held for decades around the US$2.5 billion level.

Internal IC fabrication stops being the requirement if IBM can ensure access to fabrication of its custom designed microprocessors. If GF can provide IC fabrication that IBM needs, then IBM no longer needs its internal semiconductor manufacturing capability.

If this is indeed a win-win evolution stage, what would be the benefits to GF? There are many benefits, including acquisition of IBM’s advanced SOI (Silicon on Insulator) technology and customers, and acquisition of SiGe and RF CMOS productized processes (IBM’s device models are considered the best in the industry). IBM is the source of all of GF’s advanced processing technology; it is IBM’s technology that makes the Common Platform such an increasingly invaluable brand in IC manufacturing. By manufacturing advanced microprocessors for AMD and IBM, GF would effectively preempt fabrication in that challenging segment; penetration into processor manufacturing has been one of Taiwan Semiconductor Manufacturing Company’s (TSMC’s) corporate objectives for many years. GF would also broaden its customer base and further increase its manufacturing scale. The economics of IC fabrication in advanced nodes is exceedingly harsh – being first to market and having early volume production are mandatory prerequisites for profitability.

In Summary

It is highly probable that IBM will consider a business alliance of some sort with GF. In GF, IBM has a potential partner with an infrastructure and management style that has elements inherited from Big Blue itself. With its fabrication facilities worldwide, and a foundation in complex processor design and manufacturing, GF should be able to incorporate state-of-the-art support for IBM, drive business economies, and ensure growth, noted Petrov. GF could be an ideal outlet for IBM’s IC fabrication business, enabling it to sell a business that has not met financial performance requirements for years, and still providing it the depth, scope, and resources needed to not only provide manufacturing security for IBM but also further ensuring success of its foundry

Evolution Stage Two: Acquisition of IBM’s IC Design Tool Systems and Expertise

Automation of system-level processes of any kind is the very cornerstone of IBM’s technological power; IBM’s chip design automation tools are part of this core corporate capability. In 2005 Petrov Group published a report titled “Inside IBM Research: Focusing on Design Automation and Productivity.” The report’s insights are still relevant today. IBM’s corporate DNA is to build tools for automation of processes. The results of this focus are self-evident in IBM’s Research itself – an organization that consistently outperforms its global counterparts. Petrov Group’s analysis of IBM’s innovation machine confirmed its six unique capabilities which should be of high interest to any corporation that aspires to research productivity excellence.

The three primary areas of concern to an electronic system designer (Petrov Group calls them “system survival” requirements) are power, timing, and noise. An optimal design technology would address them in an integrated manner; such a system approach is the essence as well as a unique differentiation of IBM’s chip design approach. A useful metaphor is that each IBM tool is either a leaf, or a branch, or the trunk of IBM’s design automation tree, that is, of IBM’s EDA tool ecosystem.

Such an integrated system approach is the essence of IBM’s renowned first-pass design success. IBM’s “abstraction engines,” or “the tree trunks” in the Petrov Group description, have a life cycle of 30+ years; they model basic concepts (shapes, timing, other) at such high levels that they are also used in IC-unrelated modeling (financial, materials, biological, other). As chip designs become larger and more complex, such an approach will be increasingly mandatory for successful “first-pass” design with billions of transistors in 28nm, 20nm, and finer lithography technology nodes.

IBM’s IC design focus continues to be on the needs of state-of-the-art technology. The focus has moved away from proprietary modeling and toward open systems which are mandatory for adopting third-party intellectual property (IP). Verification flow, making designs manufacturable without having to model down at the transistor level, and power and timing closure in 28nm and finer lithography all present immense new challenges. IBM’s tool systems continue to be more of a “bow wave” looking at modeling and designing at the bleeding edge and using others to maintain and support the older tools. IBM has already augmented and integrated its tool systems with industry standard tools for commodity tool solutions.

Despite its advantage in design systems, IBM has had limited success outside internal use. The external mainstream merchant market’s cost and IBM’s profitability margin requirements are too far apart. IBM’s cost structure and focus on internal requirements often make IBM the IC design partner of last resort; a customer selects and pays for IBM services only because it has nowhere else to turn. IBM provides an expensive guarantee of on-time delivery of differentiated ICs; CEOs can sleep peacefully knowing that their products will not miss holiday introduction dates.

To successfully deploy IBM’s IC design tool systems and expertise to much larger and rapidly growing segments of the consumer market, GF would have to be able take the good and differentiated and to reject the obsolete and gold-plated. This would require that GF enter this evolution stage with a clear strategy, very talented people, and continuing close cooperation with IBM. The difficult challenge will be of “mining for the nuggets” and to convert the immensely valuable technology into fiscal gold.

Summary

IBM will maintain its IC process technology leadership via research, but the critical business requirement is also that its Common Platform silicon alliance continues to be successful. The IC industry has moved away from the Computing to the Consumer sector. To be successful GF would have to meet cost economics that IBM has apparently failed to meet. This evolution stage represents an immense opportunity – if GF, jointly with IBM, is able to construct and implement a new and differentiated vision.

A key implication of GF’s and the industry’s evolution is that chip design is becoming synonymous with an industrial robotic factory. System vendors need tightly integrated chip design and wafer foundry factories. If GF is able to obtain, adapt, and cost effectively deploy IBM’s chip design capabilities it will have a decisive and sustainable competitive advantage in advanced technology nodes for its foundry customers, asserts Boris Petrov.