The Future of Foundry: Part 2

The Future of Foundry: Part 2.

Future Fab Intl.
A thought leadership project from MazikMedia, Inc.


The Future of Foundry: Part 2
by James Hines
The foundry model has enjoyed spectacular success and has resulted in a major restructuring of the semiconductor industry over the past 20 years, but it is now facing critical challenges. Foundries and their customers must adapt to a maturing industry in which the economics of technology development and manufacturing are changing.
In the first part of this two-part series, we examined the trends that are driving changes in the foundry market. In this installment, we look at how the foundry model will adapt to these changes.

Evolution of the Foundry Market
The foundry market is undergoing changes in the following three areas:

  • Increasing specialization
  • Changing investment strategies
  • Increased service offerings

Foundry Market Structure Reflects Increasing Specialization
The foundry market has begun to segment into three identifiable submarkets, and this structure is likely to persist for some time. The three foundry market segments are:

  • Leading-edge foundry
  • Commodity foundry
  • Specialty foundry

Leading-Edge Foundry
There will be a small number of large foundries focusing on leading-edge processes and designs. TSMC, the Common Platform Alliance partners (IBM, Chartered and Samsung) and possibly United Micro-electronics Corp. (UMC) will provide the most-advanced digital processes and – significantly – design services as well. In addition, these foundries will continue to utilize their fabs as they age to serve the more-mainstream technology segments. Com-petition at this level could develop into a battle between a TSMC-centric grouping and the Common Platform Alli-ance. The main customers for these top-tier foundries will be large fabless companies, such as Qualcomm, Broadcom and Marvell, and “fab-lite” IDMs, such as TI, Freescale Semiconductor and NXP.

  • Commodity Foundry

The commodity foundries will focus on low-cost manufacturing using mature process technologies. Players at this level will compete primarily on price and availability of fab capacity. Prices, and therefore margins, will fluctuate depending on the balance of supply and demand. Can-didates to emerge as leaders in this mainstream production market are Semiconductor Manufacturing International Corp. (SMIC), Dongbu HiTek and Magna-Chip Semiconductor. However, this will be a difficult market to compete in over the long term because of pricing pressures and thus low margins. This foundry market segment could evolve in a way similar to the evolution of the DRAM market, and it is one area of the foundry market that could see some consolidation.

  • Specialty Foundry

These foundries will provide niche, specialty services. Players at this level will have specific engineering capabilities that enable them to differentiate their foundry services and provide some insulation from competitive pricing pressures. Today the specialized capabilities center on analog and radio frequency devices, areas in which customers need critical support to translate designs into manufacturing. This segment of the foundry market is likely to remain small in comparison to the total market. The two leading examples of foundries positioning themselves in this segment are X-FAB Semiconductor Foundries and Jazz Semiconductor.

Changing Investment Strategies
While much attention is focused on foundries’ investment in leading-edge manufacturing processes, only 21 percent of foundry revenue in 2007 came from leading-edge processes (see Figure 1). It should be noted that the vast majority of the 90-nm-and-below segment is made up of 90nm, with 65nm representing a very small portion because it was just beginning to enter production in 2007. Half of all foundry sales are in the mainstream production process nodes of 0.13 micron to 0.18 micron.
Larger foundries, such as TSMC and UMC, have invested heavily in leading-edge capacity, but the lack of demand means that this capacity was significantly underutilized at the start of 2007. The dramatic decline in usage of leading-edge foundry capacity in late 2006 and early 2007 indicates a mismatch between foundry expectations and actual production demand. Capacity utilization recovered in the second half of 2007, due mainly to foundries holding back on their leading-edge initiatives to allow demand to catch up.

But because demand for leading-edge processes didn’t develop as expected, foundries shifted their investment focus onto mainstream, mature processes, some even utilizing 200 mm wafers, for which there is still strong demand. Many companies are still doing designs for 200 mm processes to take advantage of the less-expensive capacity that some foundries have. What’s more, this strategy is facilitated by the availability of used 200 mm equipment, which memory fabs are retiring. Foundries can repurpose this equipment at a much lower capital cost than investing in new equipment and processes.

  • Emergence of a New Foundry Investment Pattern

The changes in the foundry market previously identified could mean that a new foundry investment pattern is emerging. Instead of investing heavily in new leading-edge capacity in anticipation of future demand, foundries are now minimizing the initial investment they make in such capacity. The initial investments provide only the capacity needed to qualify the processes and support early designs and prototyping activity. However, foundries also need to invest in the design tools and third-party intellectual property (IP) necessary for leading-edge processes.

Thereafter, they will incrementally add leading-edge capacity to meet production demand as it develops. The key to making this work is to be able to respond rapidly to increases in demand. To achieve this, foundries need to collaborate closely with their customers and with equipment and material suppliers so they have a pipeline of supplies in place when a fast response is needed.

A consequence of adopting such an investment strategy for leading-edge pro-cesses is that foundries will continue to invest in mature process capacity when it’s needed. There is still sufficient demand to make this worthwhile, and the capital cost of providing such capacity is significantly lower than investing at the leading edge, which results in lower-cost capacity.

Gartner’s analysis of the capital intensity (the ratio of capital expenditure [capex] to revenue) for the foundry industry indicates that the effects of slower investment in leading-edge capacity may already be apparent (see Figure 2). This shows that the capital intensity both of foundries and the semiconductor industry as a whole is declining and that the capital intensity of the foundry industry has fallen at a faster rate and from a higher peak.

Foundries Must Look to Expand Their Range of Services
As the growth of traditional foundry revenue slackens, foundries will need to consider how they can generate new revenue streams. We recommend that foundries diversify by providing design services and packaging and test services.

  • Provide Design Services

As semiconductor process technology continues its march to ever-finer geometries (65nm and beyond), there is greater interaction between design and the manufacturing process. This provides an opportunity for large foundries, particularly TSMC, to deliver more value to their customers through advanced design services. TSMC will have to move in this direction
if it is to compete against the Common Platform Alliance, which has access to IBM’s IC and system design services group. However, it will require the addition of capabilities and development of skills that lie outside the company’s core competence. In general, leading-edge foundries will need to provide process-optimized design tools and high-end IC design services to facilitate the development of manufacturable IC designs in the future. As part of the design services, they should also expand their IP offerings.

  • Provide Packaging and Test Services

Packaging is an increasingly important part of the semiconductor value chain. To date, foundries have not been directly involved in packaging but have had close relationships with back-end subcontractors. The emergence of wafer-level packaging and advanced packaging techniques increases the value of packaging and could therefore be an attractive expansion area for foundries. It would also make sense for foundries that provide design services to provide design/test integration services.

  • Expanding Services Through Acquisition

To expand into design services or packaging and test services, foundries should consider acquiring companies with relevant expertise. We anticipate that some foundries, especially TSMC, will acquire electronic design automation (EDA) tool vendors. We also expect some large foundries to either acquire or merge with packaging contract vendors.

Conclusion
Leading-edge foundries need to embrace collaboration as a means to reduce cost and share the risk of semiconductor process technology development. Many already have adopted collaboration strategies, and these are likely to be the most successful. Those leading-edge foundries that are not actively engaged in well-defined collaborative development relationships with key partners should re-examine their strategies and consider a more-aggressive approach to collaboration.

Foundries targeting leading-edge process technologies, such as TSMC, UMC and the Common Platform Alliance, need to have a strategy to deliver advanced design services and packaging capabilities. In some cases, this might mean making targeted acquisitions to bring this capability in-house.

Fabless semiconductor companies and IDMs that are using foundry services should take advantage of the stratification of the foundry market and optimize their production loading to a combination of foundries that can provide the right mix of cost, technology and services for their specific needs.

Equipment and materials companies should recognize the ROI challenges that leading-edge foundries are facing and expect lower growth rates from this sector of their customer base than they have seen during the past 10 years. Suppliers to foundries need to find ways to add value in an era of slowing growth and increasing cost pressures.

About the Author

James Hines
James Hines is a research director at Gartner, where he is responsible for worldwide research in the semiconductor manufacturing and solar/photovoltaic markets. Prior to joining Gartner in 1997, Mr. Hines held various marketing positions at Applied Materials, Genus and Texas Instruments.

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: