Pricing of undeveloped products is one of the biggest issues for companies in high technology. They typically revolve around should one price to cost or value, which is really a marketing question which assumes profitability. Using the R&D method sets a baseline where a reasonable profit will be automatically obtained, providing the input variables are correct. Cost-based pricing often misses the target because the cost of producing a Tech good is not where the risk is. Value-based pricing is also a good benchmark. But it can be a very fuzzy number and few buyers will accept their value as the basis for your pricing. With R&D costs continually going up, this method provides a good explanation for why prices have risen as well. Plus, if you do it right, your company will be profitable, which is good for both buyers and sellers. Unprofitable products lead to unprofitable companies which ultimately go out of business, which is not good for those buyers that have invested in a product that needs supplier support. So the R&D method is an essential part in the product development tool box.
The R&D method is very simple: one first needs accurate estimates of the R&D cost it will take to get the project from concept to first revenue (Time To Money, which should be the future value) (R), what percentage of R&D your company normally allocates when it is profitable (A), the expected lifetime market potential for the product (M), and your expected market share(S). The formula is quite simple:
Price = (R/A)/(M*S)
For a real example and what it means, let’s look at lithography. It costs around $300M for each company to develop their 248nm tool sets. Divide this by 15%, which is a good rule-of-thumb for the amount typically spent by a chip equipment company on R&D. The total of $2000M represents the total revenue needed for this tool set generation to pay back its R&D costs. Now the typical annual volume of these tools was about 400 per year. One would expect to get a return after 5 years, so the expected lifetime volume to payback would be 2000 units. There were originally 5 suppliers in the market, so it would have been reasonable to expect a 20% market share. So,
(400/0.15)/(2000*.2) = 5
As it turns out $5M is the price these tools were introduced at and today they sell for more. You can run this formula on a host of markets and companies and quickly see what is right and wrong in each. For fun, let’s take a project to develop a Next Generation Lithography (NGL). Most companies estimate it will cost around $1B to develop such a system. But 300mm has dramatically increased the productivity of lithography tools so that only about 400-500 are needed each year, not the 900-1200 seen before. Plus, mix-and-match strategies have narrowed the size of the market for leading edge tools even further. So typically there are only about 60 sold each year, or a total lifetime of 300. Since only 3 competitors exist, this helps lower the price. So here is what the math says about such a system’s price:
(1000/.15)/(300*.33) = 67
Yes that’s $67M, there is no error in the decimal point. It also points to the potential problem that the industry will have to solve and that is the question of affordability.
My second rule of lithography is that the price of a lithography tool roughly doubles with each generation: 193nm scanners cost about $14M, 248nm scanners cost about $5M, last generation i-line steppers came in at $3.5M, early generation i-line systems were $1.2M, g-line was $600K, projection aligners were $300K. The first lithography tool, a camel-hair brush used to hand paint on hot wax at the mask for etching mesa transistors, cost 10 cents. Moore’s Law has functioned in this environment because of the incredible efficiency gains made by lithography tools. Imagine the size and cost of today’s chips if they had to hand paint the gates and wiring! It would be something like 30 levels of tightly spaced pinstripes covering the North American continent.
So, if you take today’s 193nm scanner that lists for $14M, one can expect that a 193nm immersion tool will be priced somewhere just under $30M. The R&D systems currently being offered are priced at $20M, so $30M for a production worthy system is not unreasonable, providing that it keeps the industry on track. The caveat is that 193-wet has to meet my first rule of lithography, which is that each generation must last three generations of device, to keep Moore’s Law profitably on track. That being the case, you get two vectors that point to an NGL price tag in the $60 to 70M range.
One might ask, why not save a bundle by having just one competitor? Three reasons: 1) with a monopoly, value pricing would dominate and the equipment supplier would own all the value. There would be no sharing of value as when there is competition. 2) Monopolies tend to be very resistant to investing in new technologies, as they expect much larger ROIs. 3) With one supplier, the chance that the wrong technology will be picked and Moore’s Law will be stopped grows. It’s the Darwinian selection model for business: narrow the gene pool of a species and the chance that a shock will push it to extinction grows exponentially.