One resource is the annual “Patent Scorecard,” produced by CHI Research in conjunction with one of my favorite magazines, Technology Review.  A summary of this year’s tallies came out in the May issue, but the entire table can be downloaded at www.technologyreview.com/scorecards/. The chart lists more than 700 companies, ranked by 2003 technological strength across eight industry categories, and compares them with their averages for 1998?2002. In addition to a semiconductor category, groupings such as chemicals, computers, and electronics include major chipmaking concerns, such as IBM, H-P, Philips, the Japanese keiretsu, and Air Products.  

Five indicators fuel the study’s analytical engine. The two most critical ones are the number of patents awarded to each company and the “current-impact index.” I spoke with Francis Narin, CHI’s president, about their research methods. They are based on an “in-house database, which is an extremely refined and analytically usable version of the U.S. Patent Office database…. Once you know how many patents a company has, the question is, are they any good? Do they have any impact?”

He says that the metric shows “how often a company’s patents from the last five years are cited in the current year. The citation of an earlier patent by a later patent is an indicator that someone is building around that innovation. Companies that have highly cited patents are those that have really clever inventions, so the current-impact index serves as a pure quality measure…of the technology.” By multiplying the number of patents by the current-impact index, one obtains the technological strength rating.

Who grabbed the top slots in the semiconductor category? Micron Technology maintained the pole position, followed by Intel, AMD, Samsung, and Applied Materials. Micron’s strength rating of 3320 resulted from its 1712 patents awarded in 2003 and its 1.94 current-impact number (which means its patents were cited almost twice as much as the average).

Although they were listed under the computer category, IBM’s numbers (strength ranking: 5386) dwarfed those of its semiconductor cousins and every other company in any category. Keeping in mind that Big Blue’s strength combines its patent portfolio, Narin sent me what his company calls a “Tech-Line Classification” for IBM. It breaks down a company’s tech indicators into 30 industrial subcategories, including  “semiconductors and electronics.” Even with these diluted numbers, IBM would rank at or near the top of the chip heap.

Narin notes that “apples to oranges” comparisons across industry-group lines don’t work, although “you can compare companies in the same technological area.” His admonishment quashed my home-team inclination to trash-talk the biotech/pharma and aerospace sectors, whose top-ranked companies’ scores wouldn’t have cracked the semiconductor top 10.

There are plenty of intriguing matchups and trends within the semiconductor category. TSMC slipped from an average rank of 7th to 10th on the current chart, joining its fellow Big Three foundries in a downward trend. UMC plummeted from 10th to 25th, while Chartered fell from 14th to 20th. UMC’s patent awards dropped dramatically, which led to its weaker ranking.

On the IDM side, Infineon cracked the top 10, rising from an average rank of 22nd to 9th place. Its patent awards nearly quadrupled in 2003 compared with previous averages. Fabless leader Broadcom almost joined the elite group, jumping from an average of 37th to 12th. Its current-impact index for 2003 was 3.29, a significantly higher number than most top-ranked companies.

While Applied Materials dominated the capital equipment manufacturer subgroup, Novellus and KLA-Tencor strengthened themselves considerably, shooting up from 36th to 17th and 60th to 23rd, respectively. Aside from its buff technological  ranking, Novellus improved its technology cycle time markedly. Narin defines this indicator as “the median age of the references contained in the company’s patent…. [It] indicates that [a company] is building its current technology on recent technology.” In Novellus’s case, it improved from 7.7 to 5.9 years, which compares favorably with Applied’s 6.6 rating.

Narin says that the size of a company’s patent collection has no bearing on its share value. “In predicting a company’s stock price, the number of patents they have never shows up as having any significance. What’s really important is the quality of the patent portfolio.”