BrainChip Holdings (ASX:BRN) September 2017, Quarterly Update Presentation

Company Presentations

BrainChip Holdings Limited (ASX:BRN) CEO Louis DiNardo provides an update on the company and discusses its recent capital raising.


Good afternoon, everyone. This is Lou DiNardo. I hope everyone that wanted to get on for the call has been able to. I'm calling, actually from the Midwest US. We've truly got a global call. Ryan Benton (BrainChip, Chief Financial Officer) is joining us, Ryan is in Silicon Valley. I'm sure some of our folks in France are on the call. Our new team member in Australia is likely on. I appreciate everybody in Sydney. It's 9:00 AM your time and for those of you that had the wherewithal in western Australia, thank you for joining us.

I'm going to make this relatively quick. I've got about 14 or 15 slides. As you know, we solicited questions in advance so that I could try to consolidate as well as incorporate those answers into the presentation. Anything that I miss in the presentation I'll try to catch up with at the later part in the call.

So, again, it's myself as president and CEO, and CFO Ryan Benton. Some of you have met Ryan. He was with me in Australia while we traveled last month, but I'll make a quick introduction when we get to that part of the presentation. Hopefully this is working well, if not, somebody please give me a ring. As I said, I'm calling in from a remote location in the Midwest.

Moving on to the second slide, this is just your standard disclaimer. Of course, you can read it at your convenience. You've probably seen it a hundred or a thousand times. And I'm going to move forward. Again, this is a September quarter update. I'm just focusing here on what occurred in the September quarter and trying to give you some color on things we talked about in previous calls. You know we make this a regular practice, so I think it's good to try and revisit the things we talked about in prior calls, as well.

The quarter was very productive for us, extremely productive in many respects. BrainChip Studio software was introduced as a commercially available product in July of this year, July 17th. That was really the amalgamation of bits and pieces of software that SpikeNet had sold as custom pieces, semi-custom pieces, to a variety of Marquee customers but was not necessarily user friendly, it wasn't necessarily scalable. We took the better part of six months, refactored the software as what you might call a shrink wrap package, a little bit like Microsoft Office, buy it off the shelf, buy your license or your key, and then off to the races. Very scalable, very user friendly. I'll show you a couple screenshots coming up in a few moments.

Later in the quarter in September, about September 10th, I believe, we introduced the hardware accelerator, which runs the same software as BrainChip Studio, but provides far greater throughput as well as density. This is again applicable for facial and pattern recognition.

During the quarter, Ryan Benton joined us as Chief Financial Officer. Originally, that was in August as the chief financial of BrainChip Inc and subsequently of BrainChip Holdings, our parent company.

Ryan and I have worked together, this is our fourth time. Ryan actually succeeded me when I retired from XR in October of 2015 and was CEO of XR, engineered a very nice exit, with the acquisition of XR by Max Linear for a little over $700 million. So, Ryan and I have worked closely and we're working together here with a team that is really reconstituting a good deal of the folks that we worked with in the past, including Bob Beachler.

Post the end of the quarter, we were selected, and I know there's some questions about this, which I'll touch on in a little bit, to provide a video analytic solution for new large scalefront climate. When we say large scale, we're talking about a deployment on the end of the user that could be several thousand cameras with a select group of those cameras which would implement the video analytics solution from BrainChip.

Again, post end of quarter, we reached an agreement with shareholders to re-escrow 403 million ordinary shares. It's actually more than came out of escrow. Many of you may be aware that on September 22nd, shares came out of escrow from the reverse takeover that was done in September of 2015, that was 361 million shares, 403 million shares went back into escrow. We'll touch on that throughout a little bit later in this presentation as well.

Post quarter, and certainly I'm sure you're all aware, we did a capital raise. Australia in $21.5 million, which is intended to support ongoing developments in AKIDA, our new amorphic processing unit, which again, we'll touch on later in the deck.

So, a very productive quarter. I think the company is probably much better situated and is well situated as we want to be right now. So, let's move on.

Again, we introduced the BrainChip Studio product in July. You can see a screenshot here. It's a very user-friendly package. It's a little tough to discriminate what these icons are, but you have an icon that says, "Learn". That's our one-shot learning capability. Take a suspect, or an image, or a pattern, one quick cursor, you'll capture that particular pattern, or person, or face. That's learn. Then you'll go with "Find" and you run your video, either streaming live or from forensic stored video, then you'll extract and report out those instances in that video stream where you found that particular subject, that pattern, or that face.

This is a very standardized software solution. It's scalable and it's applicable to a wide range of video surveillance applications. Again, that could be forensic, or it could be in real time. It doesn't require any upgrade of equipment. One of the great strengths of the spiking neural network technology is low light conditions were good, low resolution cameras were good, and this is what many of the law enforcement agencies and Homeland Security departments appreciate about our particular solution.

The infrastructure is in place. If you look at subway systems throughout the world, airports, any municipal law enforcement agency has hundreds, thousands, or tens of thousands of cameras installed and they certainly in many cases can be five, ten, fifteen, or even twenty years old.

I'm going to move forward now, touch on range of accelerator. This is the first deployment of a neuromorphic computing solution to accelerate in this case pattern recognition or facial recognition. This enables up to 16 channels of simultaneous video. If you compare that to running the software on an off the shelf server, Windows or Linux box, you'd be able to run something on the order of five simultaneous channels. This is 16 channels through a card and you can put multiple cards in a server, increasing the density, increasing the throughput. It's a 6x speed boost to the BrainChip Studio performance running on a CPU or GPU, and is 7x more efficient than a GPU accelerated deep learning solution. That is, it's much lower power, it's must faster, and you can see that in the 600 plus frames per second search capability.

So this runs the existing software, which has been refactored and is off the shelf as a user friendly interface, but it gives you more density, gives you more speed.

So, for large scale deployments where you might have hundreds of cameras, if you take for example, and we've used this example many times because we're so close to the Paris Police Force, in Paris they have 3000 cameras in downtown Paris. They've got 500 which have been described as mission critical. If you use an existing server running software with five channels per server, you would need 100 servers in order to cover those 500 mission critical cameras. If you put one or two cards and a server and you can run 16 channels per card, you can see the dramatic improvement in density. At the same time, you get the throughput, so it's really a fantastic solution for large scale deployments.

And again post quarter end, we reached an agreement with the shareholders to re-escrow 403 million ordinary shares. I think this really is a demonstration of the long-term commitment to BrainChip by the shareholders. We've got our founders over in this 403 million, we've got a couple of other large shareholders that were part of the RTO. They are long term shareholders. They're not trading in and out of the stock, and they wanted to demonstrate to other shareholders, all shareholders, their commitment to the company and re-escrow this 403 million shares. I think that's quite extraordinary. I think it's quite unusual in Australia, or any given market, for a voluntary escrow after shares have come out of an escrow.

Then post end of quarter, and I'm sure this is near and dear to everybody here on the call, we did capital raise $21.5 million, again, to support the ongoing development of AKIDA which has broad reaching general purpose applications. As we went through the evaluation of how to accomplish this fundraising, we had a number of critical factors in mind. Certainty of execution and preservation of shareholder value are really the most prominent. I spent the last month, actually four full weeks in Australia. Ryan was with me for a bit of time. Gosh, we met with probably a dozen or more different brokering firms, sought their advice, went through a selection process, and then did a deal with Shaw and Partners, which led the round. We took their advice as to how to structure the round.

We looked at everything from debt to structured finance, rights offering, and placement and determined that certainty of execution to raise the capital necessary to feel we were adequately capitalised to move through the commercialization of BrainChip Studio as well as accelerator and fulfill the promise of developing AKIDA, that $21.5 million was the number. And the certainty of execution, the preservation of shareholder value, quality of institutional investors. I think we added something over 20 institutional investors to the register.

I know there's been a lot of conversation or questions about a cornerstone investor. I think there are multiple cornerstone investors in this particular round, and I would categorize that as large checks from long term, high quality institutional investors. And by I mean high quality, the people that really do their research, are really interested in sticking with the company and are willing to take a meaningful position.

And certainlyaftermarket support, when we talk about preservation of shareholder value, it's what happens to the stock after the placement is complete. I think we've seen very nice results. We've had good volume, but we've also been able to hold up with the very nice share price. So, I think the certainty of execution, that preservation of shareholder value, the guidance that we received from our financial advisor, which is what we pay them to do for us, led us to a placement and that placement was successfully completed. I think the settlement will actually occur on Monday or Tuesday of this week. I guess it's already Monday in Australia, so Monday or Tuesday.

So, I'm trying here to answer some of the questions, consolidate and answer some of the questions that came in on the email address. I'll circle back to this when we get done if there's opportunity or time.

So, as we move forward, I think it's important to understand that our goal, our business model, is to have a very scalable and capital efficient sales model. We can use as an example, and I've done this in many meetings recently, we have a nice success at the Bordeaux Airport. There was a camera manufacturer who had installed a system that goes all the way from camera, ran the wires, trenched back to the building, put them into the server room, and installed a video analytics tool which was intended to detect perimeter intrusion, people hopping fences, people approaching airplanes from underneath that shouldn't.

This particular airport was experiencing something on the order of 1000 false alerts per month. Certainly it was costing them more to send patrols out than it did for them to acquire the system. They put our software on that server and used it as a filter on the data and dramatically reduced, over 90 per cent reduction in the false alerts, and had a very successful outcome using our software.

So the question becomes as a sales strategy, do we take that wonderful referenceable customer and do we go to Charles DuVall, do we go to Heathrow, do we go to Logan, we go to Sydney, we go to Melbourne? Or is it more efficient for us to go back to the OEM manufacturer and say, "Look, we reduced your error rate by 90 per cent. Maybe you should be installing our software with your hardware."

That OEM model gives us a dramatic improvement in leverage. We don't have to build a large direct sales force to call on every airport in the world. We go through the original equipment manufacturers, and if you could capture, in this case for example, if you could capture the top four or five vendors of cameras in this particular venue, you've got companies like Bosch, Pelco, Tyco, Exact, Hikvision, Dawa, there's five or six of them in the world that really dominate the market share. That's a much more effective sales strategy for us.

So, we could go to the camera manufacturers. We could go to the VMS suppliers. When those cameras and all those wires come back to the building, they go through a video management system, which is a fancy word for a server that consolidates all the wiring that's coming in and it allows an operator to select specific channels. It's a perfect opportunity for us to have our card inserted and bring that video analytic tool to the VMS supplier again as an OEM.

From the VMS system, it goes to a storage supplier. All this video, as I'm sure you're all well aware, all this video is stored, and that storage supplier typically has their own server architecture that sits on top of the storage element so that they can quickly, as quickly as possible depending on the media, that they can go in and do searches against the stored or archived video. So, we could go to the camera manufacturers as OEMs, the VMS manufacturers as OEMs, the storage supplier as OEMs, and that's a far more efficient sales model for us.

Additionally, you can look at system integrators, and when we talk about this opportunity that we've won in a new municipality in France, that went through a system integrator. Again, we've done a fantastic job in my view with the Paris Municipal Police Force and the French department of Homeland Security. If you take those referenceable customers and try to go to every municipal law enforcement agency in the major cities throughout the world, or do identify the major system integrators. In the case of France, for example, there's four major system integrators. We're the preferred choice for this particular video analytic with the largest, but if we can get the other three to be a preferred source as well, we cover all of France with four customers and a sales force that is really working at a business development level rather than trying to call on all of the individual law enforcement agencies, whether that be regional system integrators or national system integrators.

Certainly a third leg of the strategy is partners, valuated resellers who buy and resell product. There's kind of a gray line between valuated resellers and distributors, but distributors have a broad reach. There are a couple around the world that are the largest in the electronics distribution environment all with relationships to myself, Bob Beachler and others on our team have grown over the past decade or two.

So, this is the go to market strategy, and I think it's important to understand that this is a highly leverage sales model and one which you can get your reach out to hundreds if not thousands of customers.

Moving on, I want to spend a few minutes here, because I think there's a little bit of a semantic difference. As I get questions, I'm going to visit investors or those questions that came in online, I think the word contract is maybe a semantic difference. This is the typical technology sales process. We call it a sales funnel. We generate leads. Lead generation is probably one of the most critical and earliest parts of the process. You do that by attending trade shows, you do that with advertising, you do that with press releases. You're finding your system integrators, you're finding your end customers, the end users and who those system integrators are that they use, and you've got this mass, hopefully, of lead generation coming in as opportunities into the sales funnel.

You can see the opportunity ring and you can see a bunch of dollar signs off to the right. That's usually a very big number if you're doing your job and you've got a compelling product. From there, your sales force then qualifies those opportunities, makes sure they're real. Is it a real customer? Are they giving us real insight in to the strength of their presence in the marketplace and the compelling advantages of their solution? And usually you're kind of winnowing down from opportunities to qualified opportunities.

The thing I'd like to get our shareholders to focus on is what we call design wins. That's when a qualified opportunity goes through. We could call it a trial, we could call it an evaluation, we can call it a design cycle. But at some point, the customer comes back and says, "Okay, you're selected. You have won the design."

That's the most critical part in this process, and it's something that as our sales process matures in the next quarter or two, as our sales team gets larger and covers a broader set of customers in our chosen markets, we're going to start to report to you what our design win activity looks like. Lead generations, that's not a useful number for shareholders. Opportunities is not a useful number. Qualified opportunities turning into design wins is really where we've done our job and we have locked up our design.

Now, the rest of the funnel, design wins to revenue, is really at the feet of the customer. We don't know who's going to win in the marketplace, how fast they're going to deploy, whether they have a successful product on their end or not. Our job is to win the designs at the most qualified opportunities that we can, size that sales funnel with respect to design wins, and give our shareholders a sense of how large is that design win pipeline and how is it growing quarter over quarter, quarter over quarter over quarter. The revenue is a fallout from design wins and there's conversion ratios. That's why lots of dollar signs go to less dollar signs, those conversion ratios. For specific product lines, there's some pretty well tried and true metrics. We'll find out what those metrics are in this business as we get a broader swath of opportunities going through qualified to design wins and into revenue.

Now, the idea of a contract, I want to focus on this for a minute because in technology industry, the idea of a contract is not necessarily how we end up driving revenue. Our revenue comes from purchase orders. We're effectively a raw material supplier to some system provider. They may give us a forecast. They may give us a schedule to order and say, "I want 1000 or 10,000 units over the course of the next six or twelve months," and they'll schedule that order with terms and conditions for change orders and cancellations.

So, I think it's just a matter of semantics, but for us it's really purchase orders that count. You have your own contracts. Contracts, in many cases, for technology companies are really not a good idea. They typically don't have any commitments to volume, they're not as sound as a real purchase order, and they typically have pretty onerous terms and conditions. If you've ever seen a contract out of a company like Hewlett Packard or Apple or any of the major electronics companies in the world, it's all one sided to the buyer and not the supplier. But purchase orders and revenue falling out of design wins is where our focus is and where we start to give you folks metrics on a quarter over quarter basis as we move forward. I hope that makes life a little bit clearer.

Talking about where we're focused now with the existing branch of studio and branch of accelerator, those products are specifically designed for video analytics. As we move on and talk about AKIDA, we'll open up the opportunity for unsupervised autonomous learning as well as supervised autonomous learning, but it will have wider application than vision or video analytics. This sizes the market that we believe we address. The total available market for video analytics is about 11.2 billion as you approach 2022, and there's references as to where this data come from at the bottom of the slide.

We've carved out what we believe is about a $3 billion piece. It's $2.6 billion in surveillance, we'll call that civil surveillance, law enforcement schools, drones, satellites, driver assisted vehicles is kind of off into the, excuse me, it's not in the surveillance arena necessarily, but we put it here because it has the same attributes of vision.

Gaming, more of what we call commercial surveillance, about a $400 million market. The products that we have available to sell today have an adequate market, they have compelling advantages. We're putting the sales force in place to address these markets. I think one of the things that I did not mention at post end of quarter, we added our first executive sales resource in Australia, which is an important part of a marketplace for us. It has the same composition of civil surveillance, and frankly with gaming as well. So, I think the idea that we're now covering Europe, the United States, and Australia, and from Australia, we'll certainly reach out into Asia as well. So, about a $3 billion market, which we’ll address while we're building the next product, which is AKIDA.

Moving on to the next slide, I'm going to try and do a quick update on things we've talked about in the past, the Paris Municipal Police Force.

The Paris Municipal Police Forcewas using the bits and pieces of software that were developed by Spike Net. They certainly are looking toward the BrainChip Studio solution, which frankly just gives them a better user interface for the software that they've already been using. The BrainChip accelerator really gives them the opportunity to go to larger scale, because you can get more density through a given server. We have expectations that during the first half of 2018 we'll see a much larger deployment out of the Paris Municipal Police Force. They've given us some numbers which really are encouraging during the first half of '18.

The French National Police, or what you might call Homeland Security, they're also continuing to use the product and I think now again that BrainChip Studio is available, that'll open up a wider scale of deployment as well as the hardware which allows them to improve density as well.

Toulouse Municipal Transit, the trial's gone well, and this has been that bucket which I would call a design win where we've been told we've been selected. What they do with the roll out and how fast that goes and how large it goes remains to be seen.

Other deployments, I just mentioned the new municipality, which we put a press release out on. Just to give you some clarity on that opportunity, I call it a municipality. It's a bunch of small towns. It's got hospitals, it's got schools. They're trying to put a kind of homogenous surveillance system in place. Their intent is to deploy something on the order of 4000 cameras. There's a select group of those cameras which are, again, in that mission critical bucket. That could be hundreds, it could creep up to thousands. But I think at first deployment, we should keep our eye on hundreds.

I'm very reluctant to talk about price, price per camera for example. This is a competitive landscape and while we believe in and have proven that we've got the best solution, we don't want spoilers coming in just crashing prices with an inferior product, but we did roll that up and give shareholders a view to something that should be on the order of $1.2 million US or greater.

Education, I think we've talked about Lockport in the past. Lockport, let me back up to France for a second, because we talked about 11 initial deployments. So, if you think about a municipality that has many towns, many hospitals, they're going to bring things up in an orderly fashion, debug their system. So, they'll bring up 11, then they might bring up 20, and then they might bring up another 30, or 50, or 100, depending on how many locations this particular municipality has in their architecture.

In education, Lockport, we expect that the deployment will begin in November, the month that we're in now. Again, how long that deployment takes, they'll go school by school. They'll add cameras to a school, then they'll march on to the next school and deploy over some period of time. But that's gone well, that design win will turn into revenue. So, again, it's that sales funnel of an opportunity turning into a qualified opportunity, turning into a design win, and now turning into revenue.

Commercial surveillance, gaming, and the casino industry. We've talked about the trials that have gone on, we've talked about the deployment and Mohegan Sun Casinowhich has gone well. I'd like to think at this point that we should be in dozens of casinos on hundreds of tables. I think this goes back to scale and having the accelerator available so that we can have greater density through a server. I know with five video streams, which would be equivalent to five tables per server, it's a great number of servers if you've got two or three hundred tables. This is in an area, also, where we're focused on increasing our partnerships. As we talked about the leverage of a model where you've got a system integrator or a partner that has a big footprint in this particular select industry, that is really, I think, a primary goal for this quarter, next quarter.

We have a relationship now. Frankly, I don't think we have enough feet on the street to cover the 3200 casinos around the world, but that is certainly something that we're focused on intensely.

Cisco's been a very interesting place for us to collaborate. Again, the introduction of BrainChip Studio allows them to have a more user friendly interface and really be able to exercise the software as well as now we believe in the not too distant future, exercise the hardware. I think we mentioned before that Cisco engagement is with the Cisco Innovation Center. The Innovation Center in Western Australia, there's one here, I feel like I'm still in Sydney. There's one in Sydney. There's one in Singapore. And there's a second in Asia, all managed by the same team. So, I think we're going have great reach into Cisco, and again, scalability with the off the shelf product, I think, really makes it easier for them to populate around the world and really get a good evaluation of the product done.

Visual inspection with Saffron, that's gone well. They're starting to deploy the product. The pace of their deployment is really at their discretion, but I think you folks have either seen the demo video or at least have seen the press release. It's a very unique application for visual inspection with augmented reality and our pattern recognition tool.

So, again, from Paris Police through the French National Police Force, all the way down the list, I think this just hearkens back to the BrainChip Studio. Product was released on July 17th, so it's starting to get out into the field and populate in places where we already do business as well as new customers and the hardware accelerator for those that have interest in large scale deployment as well is also starting to gain traction.

What I didn't mention here, and we did put a press release out, which I did not mention either, was our first shipment of the accelerator card to a larger European automobile manufacturer. That's, I think it's meaningful in two ways. First, that we're entering the automotive space. We've talked about civil surveillance, we've talked about commercial surveillance, we've talked about visual inspection. The automotive space you might call commercial surveillance, and that's why we put it in a prior slide, but the ADAS, or the advanced driver assisted systems, and ultimately autonomous vehicles, will need to process enormous amounts of video. The current solution, which is BrainChip Studio or BrainChip accelerator are all supervised learning modes where you have a model, you're searching for that model as you're driving down the road. It could be a street sign, mostly street signs, but known objects.

As we move to AKIDA, which is an unsupervised mode of autonomous learning, that opens up a plethora of opportunity for us in the ADAS and autonomous vehicle space. I know there was a question, I'm trying to cover questions as I go on here. There was a question about our engagement with T2M. T2M was brought on when the company was pursuing a strategy of intellectual property sales. That is a specialty area of theirs. They're also certainly well skilled in hardware sales, and T2M did bring us that European automotive manufacturer and are shepherding that process with us.

Move on to the next slide and I think this is big news for us. This is the development of a neuromorphic processing unit. This is the culmination of all of the things that Peter has focused on in the last decade, was proven as Snap 64, which was a much more smaller architecture than this. But this is a general purpose neural processing unit, or neuromorphic processing unit. I'm sure you're all familiar with CPUs from Intel and GPUs from Invidia. Between those two companies there's a $300 billion market cap. Both large and successful companies, but frankly still using the Von Neumann architecture of serial processing, deterministic output. This is a fully parallel processing unit. It's a spiking neural network. It does do unsupervised as well as supervised learning. There was a question as to whether this would cannibalize the supervised learning mode of BrainChip Studio and BrainChip accelerator. I don't think it would cannibalize, I think it would be from a next generation equipment out of the similar customers, but I do think there's a line of demarcation with the capacity that you get out of AKIDA versus the capacity that's necessary for some of the existing applications with existing solutions.

This is, it's a big project. We're a couple of months into it. On the back of 10 years of experience. We expect to have simulations done late this year and emulate a test chip some time mid-year and by the end of '18 be well on our way to what we call hardened IC, something that we build fabulously, but we build ourselves that's a full logic implementation not done on an FPGA, very low power. We believe it can be 13 times more synapses than the competitive test chips and I'll talk about those in the bottom right hand corner of this slide in a second, and eight times more neurons. Enormous processing power. I won't call it compute power, because that's not how an NPU works, but processing power. And here we have the milestones identified as an FPGA in early '18 and dedicated silicon in later 2018.

And market applications, I think what's most interesting is it moves us beyond vision, although in autonomous vehicles, we bolt on a very, very large market to our existing market in vision systems of video analytics. But it also opens things like cyber security, as we've talked about many times, spiking neural networks are exceptional at recognizing repeating patterns. That's how vision works, that's why we're so good at it. But repeating patterns in cyber security are really the way to block DNS attacks, denial of service, block malware attacks. You can look at the Fintech market stock trading options and commodities trading, forecasting trends, intersecting those trends with your own algorithm or defining and identifying algorithms that others are trading against.

So, when you start to think about the vision applications of robotics, gaming, autonomous vehicles, drones, and air transport, and then you open up the markets for cyber security and Fintech, the AKIDA product has what we believe is somewhat close to a $5 billion served available market. You can see the XY plot in the bottom right hand corner. IBM's True North came out in 2015. It really is a test chip, not necessarily user friendly, hasn't gotten much traction. Snap 64 came out in 2016. Really proved the concept and showed proof of life for a spiking neural network. It only had 64 neurons, that's why it's called Snap 64. If you fast forward to 2018, the low-high chip that came out from Intel I guess a couple of months ago really is a test chip. It doesn't work standalone. It needs a co-processor with it to do a lot of the heavy lifting before it implements the spiking neural network. And then you can see AKIDA up in the right-hand corner again targeted for late '18. It will have all of the pre-processing elements incorporated in the IC.

That's to say when you get data and you get zeroes and ones in a data stream, IP packets, or you get the financial forecasting and stock marketing data coming in as zeroes and ones, one of the heavy lifting chores and something that we've honed as a skill is turning that binary data into spikes so that you can then process in a spiking neural network. That will be all incorporated in AKIDA and it will have significant processing power with the number of neurons and number of synapse.

So, that gives you a little bit of a sense of what AKIDA is all about. I'm going to touch on intellectual property here. I'm going to try and canvas the landscape of questions that came in. Certainly we've talked about the one foundational patent for spiking neural network, this gives you a little bit more detail on it and you can certainly go to the patent if you choose to. That patent, and one of the kind of leading indicators of the strength of a patent is how many times its cited in the competitive landscape. That foundational patent's been cited 17 times by IBM, 14 times by Qualcomm. I think a total of 35 times just in the last 18 months or two years, which means it really has a foundational footprint that others have to recognize and avoid, or come and talk to us about licensing.

We certainly talked about the JAST patents, which were licensed exclusively and perpetually. The AKIDA development is the combination of what we learned and know from the reduction to practice of Snap 64, taking the JAST learning rules and simplified neural network architecture, taking the best and combining the best of both. That is what AKIDA will eventually be. There's six patents currently pending and multiple other patents that are currently being drafted. So, we think our intellectual property protection for these large and growing markets that we carved out for ourselves and proven ourselves capable of winning are really well protected.

So, just kind of in summary, where are we going to get growth? I talked about design wins and design wins turning into revenue. We are putting the sales force in place to execute on that theme. So, we have the opportunity of growing with existing customers because we've now, we've been accepted, we've been in the field, it's been battle hardened, and maybe they want to go to larger scale and we'll use the accelerator to get them there.

New customers in our existing segments, and again, law enforcement, national security, or gaming, either going directly to end users, but more likely moving towards the system integrators, valuated resellers, and distributors that serve those end markets. Higher leverage for our sales force. And then frankly, new markets for machine vision, we've proven ourselves with Saffron in that Air Bus application that I've talked about and opportunities to grow on the back of things that we're doing with companies like Rockwell Collins, and Cisco, and others that we've touched on before.

Someone asked a question which this just reminded me of about Milipol Paris 2017, which is the conference that we're attending, I believe it's the week of November 22nd, and what we expect to get out of it? We're co-manning the booth with Rockwell Collins where they will demonstrate the strength of our technology working with their radar system. And the Toulouse Police Department, I think it's Toulouse, and maybe some of the folks from Paris as well will be manning the booth with us and showing off the BrainChip Studio software and what they do with it and how successful they've been with it.

So, our expectations out of that conference, and we were also, we're also a finalist for the video analytic solution of the year award there. So, our expectation is you've got 24, 25,000 people walking around from law enforcement agencies and security agencies around the world, and our expectation is to fill the top of that funnel with more leads, qualify those leads, and move them through design wins and take the successes we've had in this marketplace and just go bigger and broader. Again, we now have a sales presence in Australia, Europe, and the US. So, it's our time to take that next step.

The big growth opportunity on the back of AKIDA, which certainly if it comes out in late '18 would be 2019, but our goal here is to have collaboration with 8S and autonomous vehicle manufacturers, cyber security experts, as well as people in the Fintech arena collaborating with us as we architect and develop the product so that they're ready and waiting customers when the product comes out and you don't have a long time to money, or a long sales cycle, you have baked in customers when that product comes out. So, it's a patented intellectual property. It's going to carve out a very nice new market position for us, which is actually many times the size of that which we already support, and I think we've already mentioned that it's due to sample in 2018.

So, I think with that, I think I've caught most of the questions. I apologise if I missed anything significant. You can always reach me by email. You can use the same email address.

There's a question about whether we're providing our stuff for free because we've got all this activity. No, we're not providing our stuff for free. I think it's important to recognize that sometimes cash receipts aren't necessarily an indication of activity for a given quarter. At the end of a quarter, you have accounts receivable as well, and let's face it, we just released the BrainChip Studio software on July 17th and the hardware on September 10th, so I think there's a lot of activity that's going on. We should see some very nice success very near term.

I think we talked about the France surveillance roll out. AKIDA, we touched on. Again, the France contract. There's a lot of duplication here, so I'm just going to step through and see if I've missed anything big. We talked about the capital raise. There's a question about our strategy or position in terms of quantum readiness. I don't know what that means, so I can't answer it.

But look, I think during the presentation, I managed to touch on most if not all of the questions and consolidate them. Please feel free to reach out to that same email address if I wasn't clear or I missed something. But thank you all for joining and I look forward to touching base again next quarter.

Thank you very much.


Ends