Times are hard and they're getting harder. As you may recall, sometime around May 2008, MathStar went belly-up after burning through something like $200 million of investments. [These were the folks with the Arrix Family of Field Programmable Object Arrays (FPOAs)]. (Strangely enough, there's no mention of this in the "News" portion of the MathStar site; you have to wend your way through the "Transcript of the Stockholder Meeting" that you can link to from the front page).
It was sad to see MathStar go, because they had "good silicon" that had proven itself in a variety of applications. The thing that really brought MathStar down was their tools (this has been the case for a number of the more esoteric programmable logic creators).
In fact, MathStar based their original tools on Summit Designs' schematic capture suite; ergo, MathStar owed a pound of flesh to Summit (purchased by Mentor) every time they sold – or even gave away – a tool set.
Thus it was that MathStar ended up limping along with a few high-profile design wins, but their difficult-to-use and pricey tools made wide market acceptance impossible, even with their exceptionally cool, 1GHz silicon that required no timing closure.
And now we hear that Ambric bites the dust (see the related News article #212101059). As compared to MathStar, the folks at Ambric (whose website also makes no mention of their exit from the stage) had their OWN homegrown tools, based on the open Eclipse environment.
These tools were both powerful and (relatively) easy to use, to the extent that software developers could use them to configure the hardware/silicon. Furthermore, while there was a stated list price for Ambric's tools, "giving them away" or loaning them (to keycustomers) was not a problem. In fact, I heard that they were on the path to release a free version of the tools over the 'net with a self-supporting user's group – but finances killed them before this plan could be implemented.
As an aside, Ambric were absolutely on the right track with regard to their tools philosophy. The history in the reconfigurable solutions business is clear – in EVERY case where a radically new, potentially industry-changing technology has been successful, the market has been initially seeded with FREE tools. From ABEL/CUPL/PALASM and 22V10s (AMD, Cypress, and Lattice dropped 5 1/4" floppies from helicopters on potential customers) to Altera's intro of CPLDs with the original MAX tools, to Xilinx giving away copies of XACT to get customers interested in, and using, the original X2000 series FPGAs, easy-to-use and accessible tools have been the key. Ambric was almost there, and using the Internet for distribution and on-line support would have offered a HUGE advantage over their ancestors with their helicopters and 5 1/4" floppies!
But I digress... What makes the Ambric demise even more painful is that the company and its technology actually WAS on a significant revenue, design win, and overall market acceptance upswing (I hear that they had just booked TWO big-time orders on the very DAY the announcement about funding transpired! Ouch!)
Ambric was also doing particularly well in the high-end video space; several codecs for such standards as Panasonic's AVC-I, DnX (50 and 100), and Apple ProRez were already available, and more were on the way. Ambric had also been generating revenue from board-level solutions with several major video processing software companies who used Ambric's board for acceleration of the critical motion-estimation portion of MPEG-2.
The fact that Ambric appealed to BOTH hardware and software engineers significantly broadened the potential market for the devices as opposed to conventional, pure-hardware FPGA solutions. So what happened? This is where things get scary...
The way I hear it is that when Ambric's existing VCs went back to THEIR sources of money (a "cash call" as it's known), those sources were the ones that said "no way". To the best of my knowledge, this is a FIRST in the silicon business – to have the ultimate sources of cash simply write-off previous investments and walk away – especially when the company being invested-in was actually showing good progress and ramping revenues as was Ambric.
The implications of this go far beyond Ambric, and look to have a profoundly negative effect on not just the semiconductor business, but ALL businesses that have relied on venture capital to fund, promote, and reward innovation.
This is a sad day... and also a scary one... hold on to your hats folks, because it looks like we're all in for an extremely bumpy ride...
Questions? Comments? Feel free to email me – Clive "Max" Maxfield – at max@techbites.com). And, of course, if you haven't already done so, don't forget to Sign Up for our weekly Programmable Logic DesignLine Newsletter.
