At “The Workshop on Integrated Photonics High Volume Packaging” that took place in Anaheim the week of OFC 2016, Thomas L. Koch, the Dean at the College of Optical Sciences at the University of Arizona gave a refreshingly realistic presentation of the historic challenges involved with packaging chips, as well as a recitation of some limited successes, which have comprised both electronics and optics. Regrettably, his formal presentation ended rather abruptly, almost with a thud in that it was more than implied that the only hopeful solution provided would be the future work of the American Institute for Manufacturing Integrated Photonics (AIM Photonics). When Koch was asked if he could name a successful Public-Private Partnership (PPP), he mentioned SEMATECH (Semiconductor Manufacturing Technology), which could hardly be considered a triumph during the ten years in the ‘80s and ‘90s, when it was receiving US government funding. In fact, we fully expect that AIM will join a long list of worldwide PPP initiatives related to technology development, which have all failed to come even close to realizing their original goals. (We have already written about the early rumblings of discontent by some companies that have considered joining AIM.)
Koch rightly points out that the discussion of leveraging microelectronics manufacturing technology for PICs began close to a half of a century ago. Nevertheless, he still argues that the latest advancements in such packaging, as with “advanced stacking and assembly processes,” is “yet another capability that promises highly versatile intimacy between photonics and electronics [with the] potential to facilitate optical connectivity as well. Koch goes on to state that such “leveraging may be too expensive for independent development/vertically integrated providers; expenses may inhibit penetration and growth into new markets (italics inserted in last two sentences)."
Sill, after all of this time of limited progress in striving to take advantage of continuing developments with chip electronics, it is hard to justify the cost to US tax payers for the formation of AIM, especially when the country is burdened with such high levels of official debt, not including unfunded liabilities that exceed $200 trillion. In addition, there is little evidence to suggest there would be insufficient funding otherwise, such as from the federal government itself, including by the Pentagon, which would likely view this kind of development as part of a national security mandate. Of course, there is also work being conducted in the academic world like at Koch’s university, and a healthy amount of rivalry among these institutions (along with private laboratories) can help to facilitate useful new findings. Moreover, even engineers which have in recent times taken realistically negative positions on “silicon photonics” (although with a definition that has been a moving target) have had to wonder whether their doubts were justified with the tremendous amount of money spent on the technology.
A fundamental problem with AIM Photonics is that the utility of PPPs has been discredited in previous decades. The new group is not only likely to take advantage of so-called “free money” for the full five years, but the odds are good that it will request additional government outlays after that period. AIM will also by far benefit the largest vendors the most, particularly those that are paying “first-tier” dues, and these firms are obviously involved with their own separate lobbying efforts, which put them in an advantageous position over smaller players, many of which are likely to decide that internal investment would be more advantageous than in joining AIM. Most ominously, the organization could in effect result in the protection of bigger corporations from competition from those other vendors, which often because they are able to move more nimbly, frequently provide state-of-the-art improvements to the marketplace.
Much of the impact on the smaller and even some mid-sized entities will undoubtedly be determined by the details in the contracts being put together, which AIM’s CEO, Michael Liehr, acknowledges is currently consuming just about all of his time. At the start of Sematech, we know that at least some of the development agreements with semiconductor participants were forbidden from non-members for 12 months after they were generally available (eventually, this constraint was lifted). There is also the possible thorny problem of non-members (including non-American vendors) potentially forming joint development agreements with AIM members.
Another generic difficulty experienced by PPPs is sticking to a very specific goal, as Liehr had to admit recently, "At first glance a number of the proposals do not reflect as strongly as we would like a manufacturing focus (as contrasted to a technology focus)." Naturally, in targeting cutting-edge solutions, it may be impossible to neatly segment the production piece out. Although Sematech had an initially narrow scope of catching up to the Japanese markets over commodity types of chips, the consortium eventually migrated to developing production equipment and software for next-generation chips. It further moved away from its original mandate by providing cash grants to the producers of computer manufacturing gear.
Also, when it has come to PPPs, a supplier does not necessarily send the best and brightest because of apprehension about giving away too much outside the confines of the manufacturer. In addition, they follow these arcane, socialistic, and arbitrary five-year plans that inevitably result in not sufficiently keeping up with important changes in customer demand, which can occur within a matter of months in the chip space. Moreover, while Sematech was at least operating in a world market at the time with volumes in the tens of billions of dollars, how reasonable is it to believe that substantial manufacturing improvements can be accomplished when at best, production of integrated photonics is in the low single millions of units – and to what extent is it realistic to settle battles over focusing in new market areas beyond high-performance computing that may not widely materialize in the foreseeable future?
For Part II of our analysis of AIM Photonics, please click here.
[written by Mark Lutkowitz]