More responsibility for equipment suppliers?
12/01/2003
Solid State Technology asked former industry executives to comment on equipment manufacturers operating and servicing tools inside the fab.
Can equipment makers gain clout?
Daniel Queyssac, former president and COO, Front-end Operations, ASM International, Phoenix, Arizona
In the beginning, process tools were made by semiconductor manufacturers who considered them to be a major competitive advantage. It was in the early 1970s that chip manufacturers realized that making equipment was diverting resources away from their primary focus — making devices. Gradually, from the mid-1980s, chipmakers started asking for the tools and complete process flow. The competitive factor in the equation then became process integration, although the change came with loud protests about the price of tools.
 Daniel Queyssac |
To a large extent, the movement to have equipment suppliers shoulder the burden of supplying modular processes gave rise to the foundry industry and has contributed to the proliferation of today's advanced technologies. Is it possible for the transfer of functions and associated costs from the user to the tool manufacturer to continue? A number of challenges could cause a shift.
In the semiconductor capital-equipment market, volatility is a major challenge. A structure that can service a market that may expand 70% in a year, then shrink to half its size over the next three years while still able to respond to demand with the appropriate resources when the market rebounds, is not an easy feat and is definitely a major burden on profitability. Profitability is linked to two major elements: volume and/or pricing power. While it might be an exaggeration, it is not far from reality to suggest that when the IC market is flat, the equipment market goes to zero.
Volume is an elusive number in an industry that vacillates from +70% to a 40% decline in one year and where one can legitimately challenge the concept of long-term growth. Pricing power is linked to the uniqueness of a given product and/or service. In the computer segment of the electronics industry, we have seen the pricing power shift from computer manufacturers in the 1970s and 1980s to processor manufacturers in the 1980s and 1990s. Could it shift further down to the equipment manufacturer?
There are some elements present for this shift. The ever-increasing complexity and intricacies of state-of-the-art processes significantly intensify the uniqueness of the equipment, and when this is linked to the time required to switch to a new supplier for the equipment user, it would certainly point to pricing power for the incumbent capital equipment supplier.
Consolidation within the semiconductor industry will reinforce and nourish consolidation within the equipment sector and should help the equipment industry's profitability by reducing competition and choice. Fewer customers will buy higher and higher percentages of products, favoring big suppliers that have the capability to supply, ramp up, and use economies of scale for support. In the long run, however, this trend will reduce innovation because there will be less drive to compete. Development of small innovative firms will be hindered due to the high price of entry (infrastructure, global support, and organization). Those firms will be the feed stock for larger chipmakers, which will channel their manufacturing methods into their global infrastructures and subject all new ideas to their "not invented here" philosophy, stifling the innovation process.
So the question remains: can both semiconductor and equipment vendors make money and prosper?
Mutual profitability could be possible, provided that the equipment industry matures further and semiconductor manufacturers understand that helping vendors succeed is in their best interests. For example, I do not think the automobile industry would be where it is today if it had made a habit of letting customers drive their cars for free for a year before purchasing and being able to return them after the test period.
Maybe the future is for capital equipment providers to supply modular processes by the wafer, taking full responsibility for supplying the tools, operating them, and invoicing by the number of good wafers created. In such a model, it is envisioned that the IC manufacturers would decide which processes they want to use and would merely subcontract out the work to the machines on the factory floor. The IC manufacturer would still have responsibility for process integration.
For the capital equipment industry to keep innovating despite all the burdens placed upon it, the market needs to recover soon — close to its 2000 level to generate enough cash to finance innovation; and the industry needs to recoup investments made in 300mm manufacturing.
Despite the challenges, the future remains positive. Semiconductor manufacturers have underinvested for three consecutive years. Semiconductor demand is up and new technologies are coming out of the R&D labs to become sellable products.
Demand for equipment should be up significantly in 2004; the industry should recover its financial footing and, as hope springs eternal, the lessons learned during the last 40 years or so will be put to good use so the entire industry can grow profitably.
For more information, contact Daniel Queyssac at [email protected].
Equipment vendors should run production
Walter S. ("Skip") Matthews, manufacturing executive, Intel Corp., retired in 1999
The entity that best knows how to make machines perform better is the entity that designs, builds, modifies, and ships those machines — the equipment vendor. It is proposed here that equipment vendors, as a service, run wafers in production, on a price/good wafer basis, in the semiconductor manufacturer's facility.
The vendor should commit to specific, quantitatively measured, wafer-processing output parameters. If the vendor is paid on a price/wafer basis, it can improve its revenue/wafer by increasing the number of good wafers the machine can process. More wafers/hour at constant capital outlay will result in more profit. Thus, the incentive for improving the production process will rest with the equipment vendor.
 Skip Matthews |
In the 1970s and well into the early 1980s in the US, we created a conflict of interest among people who ran production. At Intel, we rewarded equipment operators with bonuses for high wafer-out numbers, and we did not penalize them for down machines. When the equipment did break down, we brought in maintenance technicians and rewarded them with bonuses based on how fast they were able to fix the broken equipment, with little or no accountability as to how long that equipment stayed on-line after it was fixed.
Through the mid-1990s and into today's production environment, we have realized major improvements in equipment utilization by moving responsibility for equipment uptime into the hands of machine operators. In most cases and for most kinds of problems, the operator is the maintenance person for that machine. Thus, we resolved the conflict-of-interest issues by making one person or one group of people responsible for both the output and general repair condition of the equipment.
In today's production environment, chipmakers systematically exclude equipment manufacturers from both production information and knowledge about specifics of equipment failures. Ironically, first-hand knowledge of breakdown phenomena is what's needed to understand how to prevent future failures. The highest priority for the equipment vendor has never been the reliability of equipment that has already been shipped and paid for. It is difficult, if not impossible, for equipment vendors to have the same level of commitment to production problems as the chip manufacturers have in the present system. Vendors do not have access to information that would allow them to upgrade their existing equipment, and there is no natural incentive for them to worry about maintenance problems. They do maintain extensive maintenance operations but these are focused, just like the ones in the fab of old, on getting problems fixed as quickly as possible.
A system in which equipment vendors are responsible for the quality and quantity of the output of their machines can benefit both the equipment manufacturer and the chipmaker.
For the vendor:
- A critical mass of technical talent could be focused on a specific processing area, and proprietary processes could be more quickly developed.
- Since vendors would own and control the process, the flow of IP to their competitors would be minimized. This could provide areas of specialization and ownership that are currently not available.
- Vendors would be able to pay close attention to the cost/layer and place ownership for that layer in the best place; essentially, the economics of manufacturing each layer will be properly focused. Integrating maintenance issues with machine design and modification capability would allow real equipment utilization and cost/process step to reach levels not possible today.
- Process steps could be aggregated and integrated much more efficiently.
- If vendors owned the equipment, they would think of ways to make that equipment reusable for future generations. They would supply leading-edge capabilities to the most advanced chip manufacturers, and then they would also be positioned to supply that same process to lagging-edge manufacturers at a later date. In many cases, the equipment that was used to supply leading-edge capability can be used again for less demanding processes.
For the semiconductor manufacturer:
- A greatly improved learning curve for integrating equipment into the process flow would result. The single biggest obstacle to getting a new process into production is the long and often painful learning curve associated with running new equipment with the new process. If manufacturers could leverage the vendor's knowledge and bypass that learning curve, they should be able to reduce the time to market of new designs and processes.
- Lower costs. Chipmakers would require lower labor costs to produce the same amount of product.
- Fewer process operations would be required, allowing manufacturing to focus incremental resources on faster process insertion, higher yields and improved process control.
Clearly there are many barriers to implementing such a concept. Perhaps the most challenging is IP ownership and protection. This could be overcome with nondisclosures and other contracts that honor proprietary processes.
Another significant barrier: Vendors would be running product that they did not own. The value of the inventory they would be processing could far exceed the value of the machinery they would be operating. Safeguards would have to be established, and heavy emphasis would have to be placed on in-line process monitoring and process control.
As the industry starts to recover, timing is good for one or more companies to consider a venture in which a true partnership is formed between the chipmaker and equipment vendors. An early adopter of a truly leading-edge process, completely run by a vendor, would be able to put that process into production much more quickly than anyone else, without increasing headcount. If history repeats itself in the semiconductor industry, the day after we take our foot off the brake, the accelerator will have to be pressed all the way to the floor. A six-month lead on the competition for a mission-critical processing step could make a huge difference.
For more information, contact Skip Matthews at [email protected].