We talked a little bit yesterday about Gelato and related applications being supported on Tesla and DP being a nice fit for that software, so is that the case?

That's more in the Quadro space of course but it's up to them if they want to support Tesla and if they do that's their software segment, so there's nothing stopping the two business units working together, but currently Tesla focuses on different applications for GPU compute.

How much of your focus is on academia, with GPU compute research, which is the other side to the market just now?

So you heard Dave Kirk talk yesterday about that and he and I work daily on those kinds of opportunities. So there's a set of applied science that even within our focus segments there's an academic chain there and Dave's research group works with that and we support him on a daily basis. That's where we're going to these places and supporting the research in terms of parallel programming and GPU computing and how you pair the two together.

And those guys are just as big a deal for Tesla and CUDA as the big oil and gas or finance guys?

Oh yeah, that side of things is just huge and they're a big customer for us and we support them just as much as anyone else. As you saw yesterday with NAMD, which is something CUDA is able to help with, they have some 20,000 licensees pretty much inside of academia, and that's just as NIH requires them to count. There's probably a lot more. So there's a whole base there that exists where CUDA is a target for us, and that's just one application. So we're actively going after these types of markets like bioscience to see where we can accelerate their root codes with CUDA.

So can you identify your largest current market for Tesla on the hardware side? We'd guess oil and gas.

Yeah, and oil and gas by need just for sea based discovery and the amount of data they collect for that that need processing, the data being collected there is running on an exponential curve that CPU processing can't keep up with. So here's an anecdote for that, so we went in to see an oil and gas customer last year, early in the year, and they had about 500TB of spinning disk that was what they used in their datacentre just for data storage for exploration, and we went back recently and that's now 2PB little more than a year later. And it's all spinning and all in use, so they have to either work faster to chew through that or they're done.

Headwave are confident there that they need GPU computing to be able to chew through those workloads otherwise they hit the wall.

Yeah, for exploration there's a need there to be able to work on data in two parts, both in the pre-drill simulation and then also as they drill to make sure the 150 million dollar hole is working out as they expect, and that needs to be done pretty much in real-time. If you look at Exxon-Mobil and they're a 400 billion dollar company and say the top 7 in oil and gas is a 3 trillion dollar market, so it's that scale of business and their computing needs match their business sizes there, so there's a lot of interest by them in Tesla.

so companies like that have a need to stay on the bleeding edge in terms of compute power for their money, and the GPU is something they're looking to, to help there?

They always use the next thing, since they're business is built on how much data they can process. If they can potentially double the amount of exploration data they can process in a given time, they can double the number of holes they dig. So their compute power can dictate how their business grows.