End of the road for MathStar (for real this time)

1/13/2010 3:35 PM EST

If you have trouble seeing the connection between programmable logic and translation software, you are not alone.

MathStar has been for months something of a phantom company, reportedly with just one employee. Calls to the company's headquarters don't reach live people (though it is possible to scroll through the still active voice mail boxes of dozens of former employees, like ghostly voices on an abandoned ship). The company's website is still up, but apparently hasn't been updated in a long time (it still lists as its lone employee Chairman and CEO Douglas M. Pihl, who reportedly resigned over the summer after the company signed a non-binding agreement to buy Sajan.)

The bottom line is that, as reported by the Minneapolis Star-Tribune, the Sajan deal spells the end of the line for MathStar, officially. MathStar will take its cash reserves of nearly $13 million (more like $7 million after the acquisition) and head to Wisconsin to work on language translation software. The combined company will operate under the Sajan name.

MathStar threw in the towel in 2008 due to slow sales of its field programmable object arrays (FPOA) and the fact that no buyers were forthcoming. At the time, the company said it halted development in order to consider strategic alternatives for its assets. Not sure if anyone anticipated a move this drastic, but the company's board obviously felt that Sajan represents a good investment.

As for what will become of the FPOA technology, it seems as though it will simply be discarded by the wayside. From the jump, MathStar's architecture represented a fundamental change, with silicon "objects" in the form of 16-bit arithmetic-logic units, each with its own instruction cache and scratchpad memory. The history of PLDs is littered with failed startups that tried to do something that was too fundamentally different for users to embrace, even if it promised certain advantages.

Some shareholders of MathStar had pushed hard for the company to liquidate its assets and return the proceeds to shareholders. But that plan of action was apparently voted down by shareholders. You've got to figure that some of them are not too thrilled about jumping into bed with Sajan. Apparently Pihl was one of them.

It's also interesting to note that MathStar spurned several takeover offers before choosing this path. But those offers don't appear to have adequately valued the company, considering the cash on its balance sheet.