HPC, market appeal, and stock prices

Print Friendly, PDF & Email

The overall tech stock market has been slightly up this year, but supercomputing pure-plays Cray and SGI are significantly down. Current shareholders may bemoan this schism and wonder why other investors are indifferent about this field. One opinion is that the high-performance technical computing market is so niche that it doesn’t generate enough buzz in the broader media.

For example, Mellanox and Voltaire have had dim IPOs, while VMware’s has been on fire. EMC was able to make a spin-off that captured broad appeal by marketing their product to enterprise customers: VMware aims more for ERP and CRM than CFD.

Likewise, compare the stock performance of IBM and Sun to that of HP over the past few years. Only one of these has not been chasing DARPA’s High-Productivity Computing Systems program. Similarly, QLogic and Cisco have both entered the InfiniBand market through acquisitions, but only Cisco’s stock price has consistently performed well since the merger (Cisco doesn’t brag about their MPI latency).

However, Google’s recent acquisition of PeakStream did make some mainstream tech publications. The consensus on Google’s strategy was that better performance and productivity will be required in future enterprise-class systems. Indeed, this very move may signal how traditional HPC professionals may be able to capture broader appeal by applying their craft beyond technical applications.

The HPC community may do well to use their collective knowledge to muscle into the data center. An existing company may be able to “break-in” by “selling-out.” That is, a small technical-focused vendor can merge with a larger enterprise vendor, but only if the strategy is to use the intellectual property to further the goals of enterprise computing.

Another approach is to retool the firm as an embedded systems player, though this strategy isn’t as widely used. However, this approach does seem feasible if it opens the company to a broader base of customers.

One word of caution though is that blindly becoming an enterprise or embedded systems player is not a guarantee of success. The goal is to ultimately increase the customer base. A data-center company with a small customer base will also scare off investors, ala Rackable Systems.

In conclusion, broader appeal usually translates to a better stock price. Given that high-performance technical computing is such a niche area, many professionals in this space would do well to investigate applications beyond their existing domain.