This provacative and iconic statement made by AMD founder Jerry Sanders in the 1980's, was intended to drive a stake in the ground about competitive advantage in the semiconductor industry. The 1980's were a time of tremendous turmoil in the industry. The low cost of Japanese capital fueled a fab boom in Japan that threatened the existence of the US semiconductor industry. The American industry had gone through several boom and bust cycles, and the Japanese industrial base was threatening the American base. DRAMs were showing some signs of maturity, and some were calling manufacturing a commodity skill that was best done in Asia. Others were arguing the opposite. Jerry Sanders was clear where he and his company stood.
Despite the massive changes of the ensuing 25 years since Jerry Sanders made the statement, it is still quoted, although often in the context of a counter-argument along the lines of "Fabs in and of themselves are not strategic, business models are." The fact remains, however, that the fabless or foundry model works well for mature process flows where competitive advantage lies in chip design and application. On the other hand, it resonates where the manufacturing process is highly specific, complex, and enables a differentiated product. Digital foundries are highly successful because they execute well-defined design building blocks using standard process flows on standard wafers. Are there any analog foundries?
"Richardson, Texas, October 18, 2012. Photodigm announces the availability of the world's highest power spectroscopic lasers designed for Rb and Cs atom sensors. These new high power DBR laser devices have powers at least 50% higher than Photodigm's previous products, and are the result of substantial advances in its laser design and fabrication capabilities. As with all of Photodigm's DBR lasers, these devices have single longitudinal and single spatial mode beams. "
Now what is it about a fab that has enabled these devices? Our devices are fabricated with 6 mask levels and up to 120 individual process steps. Every detail is important, beginning with the epitaxial material --from initial design of the epi, the characterization of the epi as fabricated, and the tweaking of the device design to match actual epi. Critical structures must be etched precisely to select and guide the fundamental spatial mode. Subwavelength gratings must be fabricated to specific dimensions with tight tolerances to select a single longitudinal mode. Multiple other features must be carefully fabricated to preserve the mode characteristics. Substantial intellectual property resides in this fab.
This attention to detail cannot be outsourced. It becomes the core of the business, the primary differentiating characteristic of a Photodigm Spectroscopy Certified™ laser.
The result is a device capable of delivering a single spatial and longitudinal mode, with a linewidth of about 1 MHz, capable resolving the hyperfine structure of the Rb or Cs atom, with a power output undreamed of even a few years ago. Furthermore, advances in performance are accompanied by advances in reliability and lifetime.
These new high power devices, now shipping in volume to select customers, are the result of substantial development efforts supported by multiple partners over the years. The work was funded by SBIR contracts from National Science Foundation, the United States Air Force, DARPA, and the Missile Defense Agency. While the current customers are primarily research and development focused, atom sensors using laser oscillators locked to alkali atom transitions show great promise for advanced capabilities in timekeeping, navigation, and timing. Photodigm's efforts are continuing, and our customers can expect to see increasing power levels and improved reliability in the upcoming months and years.
At Photodigm we take the competitive advantage of our fab seriously. You can't just buy commodity Fabry Perot lasers and dress them up with a hybrid element like a volume Bragg grating and make devices of the level of performance that our customers have come to expect. You can't take an external grating and a commodity laser chip and make a device with the ruggedness, stability, and compactness of a monolithic DBR chip.
It is tough, but our fab team of Duy, Rick, Lan, and Chris, working with our design team, are real experts delivering real products out of our fab.