BrainChip Holdings (ASX:BRN) Company Update Presentation, December 2017

Company Presentations

BrainChip Holdings Limited (ASX:BRN) CEO, Louis DiNardo provides a year-end review for 2017, talks about sales process and funding, the company's Akida platform, opportunities and what’s planned for 2018.

Good morning everyone, thank you for joining the call. This is more of a yearend update than it is a fourth quarter update, as you're all aware, we don't close our quarter until the end of December, and we'll get back to you sometime after the 4C goes out, I guess that'd be late in January. Again, thank you for joining the call, I'm here in our San Francisco office with our CFO, Ryan Benson, I'm going to see if I can get the presentation up. There's not a lot of feedback in this type of system, so I'm hoping that the presentation is up and live, and I've got about 10 slides. I will say before we start that I did take the time to read all of the inbound questions, and have tried to incorporate as much as I could in the way of answers as appropriate into the deck. I also had one kind of funny input that I speak too fast and sometimes I'm difficult to understand, so I will try to slow down and make sure that I articulate well.

Again, $32.9 million in three rounds, all done by private placements, we've talked about the reasons why, the advice we received from our brokering firms. The institutional register is really quite healthy, marquee names, 30 new investors or more across those three rounds, you have institutional investors. Maybe notably as well, our board is now US-centric, Julie H. Stein joined in the later part of '16, has been with us now almost a year and a half, of course I joined the board in November, about a month after I joined the company, and then Manny Hernandez was our most recent addition. We did just recently announce that Peter van der Madehas committed 100 per cent of his time to the Akita development.

Peter's passionate about that development, wants to see it not just be successful, but that we can accelerate as quickly as possible development and introduction of the Akita, what we're now calling neural system on a chip. We did announce in Peter's resignation that we will be starting a search, have started a search for one and possibly two directors. We're authorized for seven, we currently have five. One of the questions that I received is what are the competencies that we might look for in those directors. You we have a diverse board as it is, and Julie's background, Goldman Sachs, Solomon Brothers, building a large investment firm herself with her partners, good strategic thinker, adds a lot to the board in the way of governance as well as strategic planning.

We know I'm not an executive director, so I won't go into that, but Emmanuel T. Hernandez, probably one of the marquee technology directors in Silicon Valley, comes with very strong financial background, he was the chief financial officer at Cyprus Semiconductor and at Sun Power, sits on several large public company boards here in the US. Of course you know Mick Balto and Adam Osseiran our Australian director, so that we maintain those two residents that are required by ASX listing rules. I think with respect to new directorships, you know we really want to have the best diversity and the most appropriate skillset. I think in one case, it would be nice if we could find a seasoned director with public company experience that comes out of one of our target end markets, and whether that be cyber security, financial technology, autonomous vehicles, or surveillance, the existing product line service in surveillance.

I think that would be a nice way to round out the board with some real good, solid direct industry experience, and then maybe another director who really has a very technical background, someone who's been in the process of business, you know could come out of an Intel, an IBM, an Nvidia, just to give three names of companies that people know well. That's not necessarily where you hunt for directors, but just to give you some sense of what that skillset would be. Also during the year, we rounded out our executive staff, Ryan joined us, Bob Beachler'both folks that I've worked with a long time, Ryan is a four-time CFO with me, Bob Beechler was also vice president of marketing at XR Corporation, where I was CEO, and then followed by Ryan as CEO.
 Tom Stengel joined to run sales in North America, Tom worked with us at XR, Greg Ryan joined us in Australia, because having some feet on the street in Australia is very important for the relationships that we've developed, and really closing in on design wins and revenue.

We will add someone very very quickly here, I'd like to make an announcement maybe as early as next week, a sales executive in Europe, outside of France, someone that can really leverage their experience in the UK, Germany, Spain, as well as French markets, so that'll be an announcement quick to come. We talked recently the last call about feet on the street, this is a business that requires a sales presence.

In the US, we tend to use manufacturer's reps, so we're engaged with a large manufacturer's rep located out of Los Angeles that covers the entire southwestern US, it's a 25 or 30-year-old company, I've done business with them for the better part of 20 years. That would put a few dozen people on the street on our behalf, also selling other lines, that give them untrained customers. It's a pretty tried and well-worn path in the US, is what we call super reps that cover the southwest, the northwest, the central US, as well as the northeast, and then the southeast as well. That really adds leverage to the sales function, when you can leverage 10, 20, 30 sales representatives, all tend to be EEs, all tend to be very highly skilled business development as well as technical salespeople.

Well capitalized solid board of directors with a desire to add more talent to the board, as well as a solid executive sales team.
There was a question about our head count, our head count currently sits at about 27, that's about five people in sales, and that's sales and applications engineering, but as I just described, more feet on the street coming very quickly, three people in GNA, we know the CEO, CFO, as well as another finance person in France, and then about four people in research and development that work directly with Peter on the Akita architecture. The balance is about 15 or so, 16 actually, are hardcore engineering in the development ranks. We tend to be split about evenly in the engineering and research teams, about half in Toulouse, France, about half in Newport Beach.

We'll move onto the next slide, continue the year in review. I think one of the most important things we accomplished very late in '16, very early in '17, was to reset the markets that we addressed, the product and technology strategy. When I joined in October of '16, I guess it was really September 29th, we'll call it October '16, as we've talked about before, we had just acquired Spikenet. They had a lot of custom pieces of software, a marquee customer base, which I found very advantageous to us as we moved to commercialize that software, refactor the software, and introduce what you now know as Brain Chip Studio. We set our sights on the FPGA to work in concert with that software, which we now know is Brain Chip Accelerator.

It's recognized that BrainChip Studio and BrainChip Accelerator are specifically designed for vision applications, whether it's machine visioned, whether that is surveillance, so the markets that we're addressing, the revenue that we'll generate in 2018 and probably the first half of 2019 will all be on the back of vision applications, and we'll talk about some customers and bring you up to date on the status of some of the deals that we've talked about.

As you know, Studio was introduced in July of 2017, Accelerator was introduced in September of 2017. There's been questions about how much activity have we seen, I can tell you there have been dozens of software drops on Brain Chip Studio.
The product is being demoed in the hands of dozens of customers right now, Accelerator, the hardware solution, which was released in September, we know they did a press release, the first shipment to an automotive manufacturer who has now had the card up and running for about a month or so, so I'm sure there's a lot of good work going on there, and our engineering team is in close contact. We have an RFQ in the house from I'd say one of the world's largest networking manufacturers, different application, we're digging into what it is that they would like to accomplish with Accelerator, some of it is to learn the technology, and some of it may address specific applications.

Given July and September, I think we're really pleased with the activity that we've had. I'm going to touch a little bit here on major customer activity, and then follow up with that in a little bit more detail. Major customer activity really breaks down into four or five categories. We've got our gaming customers, we're now engaged with VMS suppliers, VMS is video management system companies. When cameras are hung, wires come back to an enterprise or a data center, there's a VMS system which all the cables connect to, that's where the operator really controls his video system. There are two notable VMS suppliers in the world, there's many, but I think the two most notable are Monostone and Gentech.

We're making progress on both fronts, we're engaged with a storage vendor. You could think of storage as where does all that video go once it goes through the VMS system? Just to throw out some names, companies like Dell, EMC, Western Digital, Quantum Corporation, these are large providers of storage for video surveillance, and they certainly would like to add more capacity or capability to do forensic search on the stored videos that they have, and provide that as a service to their customers.
Cloud services, cloud services are really analytics online, software as a service, and you'll see that I'll note this later, we're talking to companies in advertising, talking to companies in retail for consumer behavior.

Those cloud services are kind of a pay as you go model, so whether they drop Brain Chip Studio in, where they have so much demand that they want to drop in Brain Chip Studio running on Accelerator, that would happen in the cloud and they'd pay by the bit or video processed. I'll make sure I'm not missing any questions as I go along. A little update, and really not a whole lot to update on Saffron, Rockwell Collins and Cisco, all are using the product, you know we're kind of at the discretion of them on how aggressively they deploy and how aggressively they market the product. I think you are all aware that they presented with us or did booth duty with us at the recent Millipoleconference.

Saffron is deploying the vision system that I think you're all well aware, but I think we've actually had it in a couple of presentations now, and I think we'll benefit greatly with Cisco now that we have Greg Ryan on the ground and we'll have some local engineering resources as well soon.

Cisco is focused on several different applications, including vehicle identification, facial recognition, a wide variety of things that could be done with either Brian Chip Studio running on their own hardware or Brain Chip Studio running on the Accelerator in one of their boxes. All in all, things are going nicely with Saffron and Rockwell Collins, Cisco and others that we've talked about, we'll keep you up to date, and probably give you a little bit more of an update when we do our first quarter call.

Along the same thing, you know moving away from the year in review and looking at really what's going on as we develop our approach to building a sales funnel, so I'm now on slide five, you can see the lead generation process for us is conferences, advertising, press releases, maybe sometimes more importantly than press releases is press coverage. I know there is a lot of focus particularly in Australia, as we're a listed company there, on ASX releases, we did 29 ASX releases in 2018, we do our best to keep our shareholders up to date, we use ASX as appropriate. Sometimes what we think is appropriate, they don't think is appropriate.

Take for example when we won the award at Millipole, we certainly wanted to get that in the press, both generally through editorials and press coverage, but the ASX found it to be a marketing piece and not something that was appropriate for the ASX. Bear with us as we deal with ASX and what's appropriate in their minds and what's not, but if we can't get it on ASX, you'll find us releasing it through trade magazines, getting into financial press as well, AFR really likes to pick up stories on the company. Another question came up, and you can see the two logos off to the right about Millipole, and I'll just touch quickly again on security today.

Security today generated a lot of lead generation, those leads then turn into opportunities. As we qualify them, they turn into qualified opportunities, and that's where we are in a lot of that process with security today inputs. We generated something over 80 leads out of the Millipole conference, I think that was a really healthy response. I know having Rockwell Collins in our booth with us, having the Paris Police Force in our booth with us doing live demonstrations of how they use the product really resonated with other law enforcement agencies. We're following up on those leads, qualifying those opportunities, hopefully those opportunities into design wins, and of course then design wins turn into revenue.

Then there was a question about how is this sales funnel starting to build up. We'll say that as we come out of the fourth quarter and we speak during the first quarter, which again will probably be sometime late January, after the 4C is out, I'd like to be in a position to put some numbers around these things. We're bringing up a new CRM system, qualifying which of the platforms we'd like to use. is probably the most prevalent, also happens to be most expensive, and there are lots of alternatives that will give us a real view of what the sales funnel is, taking it out of an Excel spreadsheet and putting into a real CRM platform.
Our design lens, I think given that the product has really only been released since September and July, but there were certainly a backlog of folks that were using the more custom versions of the software, I think that you can expect that the numbers we're going to talk about again in the late January early February time frame are going to be design wins that are approaching maybe $10 million the first year, deployment and something on the order of a lifetime value for those design wins that could be maybe as much as $50 million or more.

Could be much more than that, as we scrub the funnel regeneration and opportunities. How do we qualify a design win is a big part of scrubbing the database and making sure the management really is persistent in making sure that the design is really locked down. At that point, turning into revenue is really up to our customers, and we state quotes and we'll keep you folks as well informed as we can based on the information that we get.

If I move to slide six, and this is covering things that we've all talked about in the past, civil surveillance as well as commercial surveillance. In the school arena, we've talked openly about Lock Fort.

I think last time I spoke at the end of last quarter, we said that Lock Fort hadstarted their deployment, they did in fact start their deployment in November. I think we got a pleasant piece of news just recently, Lock Fort had decided to launch that development, still waiting on full funding, but I forget if it's a bond or some other government mechanism, but they were going to fund themselves, it was going to be a slower process, but that funding has come through and now the deployment has moved much more quickly. We're still also certainly in discussions with many other school districts, principally in the state of New York now, because once you win one and you're qualified, it's a whole lot easier to knock down the others.

Municipal law enforcement is going well, and expansion in Europe is important to us. I'll touch on what's going on in Paris and other municipalities, but expansion in Europe is important to us, and that's why we're adding a new sales exec in Europe, to focus not simply on France, but take us into Germany, take us into the UK, Spain, even over into Turkey.

The Paris Police Force, things are moving well, and I think there are plans in the first quarter to deploy more cameras as well in the department of Homeland Security, you know the French version of the department of Homeland Security, things are going quite well, and I think that we'll see growth in the first quarter as well.

I covered schools, municipal law enforcement, Homeland Security, and as I said, Paris is doing well, we talked about another deployment in municipality, that's going well, really expect it to start in earnest in the first quarter of 2018. What we call commercial surveillance, well actually I'm sorry, I shouldn't skip over airports because there was a question about Rockwell Collins.

Our work with airports is really a work in progress. Now that we have a stronger and a broader sales team, more airports can be covered, or the system integrators, most importantly, who work with those airports and OEMs such as Rockwell Collins. Rockwell Collins, they are out marketing their infrared and radar system, you know we'll see both Brain Chip Studio as well as Brain Chip Accelerator opportunities in the airport environment.

Commercial surveillance, gaming, retail, advertising, traffic management, and automotive, and I'll touch on gaming first. It's really a matter of sales coverage, we talked about this last time, feet on the street. We've had some significant developments with a potential OEM partner, you know the SN Tech relationship, which I can touch on again in a little bit, but the SN Tech relationship has been in existence for a couple of years. I think they are resource starved, so we're working very closely now with one of the largest OEM providers to the gaming industry, about 3,200 casinos in the world, they do business with all 3,200 casinos. They have Sales Force, they have a service organisation, would be a real win for us to have that large and broad relationship, so that's I hope very near term.

Retail, again retail and advertising is more likely to be cloud-based, where we work with a partner who has a cloud-based system up, people rent time in the cloud and they define, "There's logos that I want to see, here's behavioral patterns with consumers that I'd like to see." Now we've got a nice burgeoning relationship with one of the most notable SAAS suppliers in the advertising and retail arena.

Traffic management is really, at this point for us, it's Cisco, we've talked about that previously. Now that we have Brain Chip Studio, which is a commercial user friendly product, I think with Greg's support down on the ground in Australia as well as some engineering resources, that can move quite quickly.

Automotive is autonomous vehicles, primarily ADAS or advanced driver-assisted systems, there are opportunities for Brain Chip Studio and Brain Chip Accelerator. As we've mentioned, you know we shipped our very first Brain Chip Accelerator card into a large European automobile manufacturer, you know for a wide variety of venues, and that could be the publicity that we got on shipping the cards, it could be the publicity that we've gotten at trade conferences, but we're working with a wide variety of automotive manufacturers or third party manufacturers, both to get Brain Chip Studio and Brain Chip Accelerator designed in, but maybe more importantly to work with them as partners in the Akita development, so that as we develop Akita, we know that it's well-tailored for these applications, we know that there's a customer waiting when we get to the end of the rainbow and the product is launched.

I'm talking about companies that we're having dialog with now, including Volkswagen, Renault, General Motors, you know the list goes on. It's really a matter of intercepting them at a point in their research and development where they can have influence on Akita or they can leverage Studio or the Accelerator.

There was a question about Tesla, and what we might be able to do with Tesla. I know some of you recognised an article that Tesla has dropped Nvidia in their autonomous vehicle program. I think it's important to separate what goes on in the autonomous vehicle, and even when we talk about cyber security or we talk about financial technology.

Spiking neural network is exceptional at recognizing repeating patterns, GPUs and CPUs are really workhorse computational engines, they do math. We don't do math, we don't do computation. If you talk about Fin Tech, and I'm going to jump ahead a little bit because I'm on a little bit of a theme here, when you talk about Fin Tech for examples, a lot of questions about cryptocurrency. A spiking neural network will never be the appropriate solution to calculate a transaction, one plus one equals two, should this transaction clear?

What the spiking neural network in Akita will be exceptional at, whether it's for autonomous vehicles partnered either with GPU or CPU doing calculations or in cryptocurrency or any financial technology, high frequency trading, options and commodities trading, is not doing the transaction, it's looking at trends in data that you might want to trade against or you might want to use to make your autonomous vehicle smarter over time.

Or in the cyber security arena, there's really not a whole bunch of computation that's necessary there, but repeating patterns for denial of service attacks, malware attacks. Let's separate when we talk about automotive, and for example, an opportunity with could be Tesla, could be Uber, could be any one of the big names that are moving into AV or autonomous vehicles. We'll be there to help their system learn, there will always be a CPU or a GPU doing calculations, and that will be similar in financial technology, particularly here if you talk about crypto or even blockchain technology even more generally. I just wanted to wrap those things up so that people understand Akita, its ability to learn autonomously repeating patterns so that people can look at trend data, make decisions very quickly, that's really where we'll fit.I'll touch on that a little bit more when we get to the slide on Akita.

Slide six, we're going to move to slide seven, which jumps right into Akita. Again, a neuromorphic system on a chip, I'm going to go to the bottom right hand corner first, and why we've decided to move from the idea of calling it a general processor more so than a system on a chip is because it really is a system on a chip. If you think about the last time I presented here, we talked about Intel's introduction of a research chip called Low High, at 131,000 neurons and a couple of hundred million synapse, but it lacked all of the pieces that were necessary to get to the neural network.

There are a whole bunch of blocks that are necessary, transducers in the world, things that pick up information, if it's real world data, sights, sounds, pressure, temperature, don't output sights, they output zeroes and ones out of data streams. One of the strengths of what we've done in the FPGA that's implemented in the Accelerator is we have a block that is a Spike converter. We take inputs that are not spikes, and we have some very sophisticated approaches, some which are wrapped up in patents, some which are wrapped up in trade secrets, that convert that data into spikes. If you have a standard camera, you're getting pixel data, if you have a spiking camera, great, you can bypass the pre-processing, but for data inputs, if it's data, financial transactions or cyber security, that data has to be converted into spikes.

It's a very sophisticated process, something that Peter and the research guys have spent a lot of time and had great results with, and Neil and the development team has been able to implement a real proof of life in the existing FPGA.

You go through the spike converter taking the inputs, you've got communications interfaces, you've got power management. It's kind of like what we're doing now is the floor plan of a home, you know you've got six bedrooms, you've got a living room and a dining room, you've got hallways and a garage, where do all the pieces go, how do they best interface to other pieces within the framework of that floor plan, and how do you get in and out?

That's where we are now, I'll talk a little bit more, I guess I should just talk too on this slide, there's questions about the timeline for the Akita development.

I think we're really in good shape, Peter and the research team have done a great job in defining the architecture of the neural network, the blocks that are necessary to build this system on chip. There's been a lot of simulation done in what's called Mat Lab, there's simulation and development that's going on in what we call C++, which is a programming language. We expect to have a C++ kernel of the neural network maybe by the end of the year, maybe the first couple weeks of January. One of the questions, let me finish that thought. From there, now you've got something that can actually simulate the process of taking inputs, again, pixel camera or data for cyber data, data for Fin Tech, you can run an algorithm against it and you can prove that that network is robust, that it works as planned.

The FPGA is still planned for mid-year, that's third week of June, first or second week of July, or probably be June, early July. That FPGA will implement all of the functionality, so it won't just be exercising the kernel, it will be exercising the pre-processing block, so it'll be exercising probably not necessarily the power management blocks, because when you get to the ASIC which has many multiples of the gate counts, your power management scheme will change, but it will exercise the functional blocks of taking the spikes, converting them, doing all the things that need to be done to get the neural network to operate.

The FPGA is expected mid-year, and the ASIC development from there, the process, I wouldn't call it easy, that would be disparaging to the hard-working engineers we put in, but your digital design is essentially proven.

Most of the blocks, and what gives us high confidence that this is a risk reduced effort, is most of the blocks, for example the spike converter, many of the communications interfaces are already implemented in the existing FPGA. If the spiking neural network design in the C++ kernel is debugged and simulated well, it's a very much risk reduced design. That would lead us to probably launching into what we call hardened silicone, call it an ASIC or hardened silicone sometime late in the third quarter, maybe have place and route, which is the physical design done, so that we can launch wafers into a fab and have something late in '18, early in '19. That's the timeline that we've been talking about, there's no reason to believe that that timeline is changed, and in fact I think we've got reinforcement that it's still well within the realm of us accomplishing it successfully.

Just a little bit more on this page, I probably should have put this page in here, but there was a question about our IP portfolio. I touch on it a little bit in the product description, but just to be a little bit more clear, we do have the one foundational patent done, spiking neural networks, that was issued in 2008. It's been cited many many times, I think last count was 37 times, IBM, Qualcomm, Samsung, anybody that wants to play in this space recognizes that patent as a foundational patent and cites it.

We've got five additional patents in process, several being written around things that are being developed with and for Akita. Let's not forget that we own exclusively and perpetually the JAST license patents.

I think we've got a pretty good fence around our IP, we've got a dedicated team having the most robust strategy that we can. IP is really what the underlying value of this company is, you execute well on delivering that IP, compelling and differentiated advantages for customers, and that's how you win. I think I talked about most of what was on this slide the last time we spoke, we've got the development milestones, the end markets, I touched on autonomous vehicles. In vision systems, whether you'll need Akita or not, whether the Accelerator is adequate, there was a question which I'll kind of reframe about Akita, which could never be done with the Accelerator might be able to be implemented with Studio, but frankly wouldn't be big enough or bold enough, and that is, can Akita work at what we call the edge?

So far what we've talked about with Accelerator is all enterprise class, it's network oriented, it's a data center, it's in an enterprise.
Can Akita find its way at the edge into, I'll just use the basic example, can it find its way into a camera such that the camera is actually performing the analytics, and only transmitting back the required information when there is a security alert or when there is a behavioral pattern, when there is an image that's recognized?

Akita is going to be very scalable, it's not going to be a terribly large IC, it will be extremely low power, and whether the end users or the OEMs would use the device or whether they'd come back and say, "We're building our own very large IC, now we'd like to license your IP," which kind of circles all the way back to part of the business model when the company was founded back in 2015.

I think it's been prudent that we've moved down a course of proof of life, total solutions so that customers can see how things work and what the advantages are, what the power performance is. The Akita solution certainly could be an edge capable device.
Make sure I don't miss any questions there. There was a question about early engagements with Akita. While we're building our business around vision applications and surveillance applications, and we can build a very fine business, very profitable, high growth business with Studio and Accelerator and enhancements that we're doing to Studio and Accelerator. Early engagements, again, they'll be with some of those leading autonomous vehicle or automotive manufacturers, they'll be with leading cyber security companies, they'll be with leading Fin Tech companies.

The benefit of having a kernel that we can run algorithms against is that we can show how what we're working with the benchmarks of performance are, so that's why we need to have collaborations and partnerships very very early in the process here, those are the things we're working on now.

As the FPGA comes up, it will spawn a development system, so people can actually start to develop on the Akita platform, on the FPGA and the development tools that go with it. Now some of that will be oriented towards industry, and some of it will be oriented towards data scientists. Now there's a great wave of data science that I wouldn't call it research, but it's in the academic realm, it's within research labs that are within large Fortune 500, 200 and 100 companies, but there's a lot of innovation that comes out of there.

That development system will be early engagements for both real life industrial customers, as well as data scientists. I think that covers what I wanted to talk about with respect to Akita.

There's a lot of questions, I did read everybody's questions, I can assure you that, there's a lot of questions about 2018 revenue, when do we get revenue traction? Again, it's all about the sales funnel. We've got great design rooms, we do need more feet on the street, particularly in gaming, we need the collaboration with cyber as well as autonomous vehicle and Fin Tech. We don't publish a forecast, you know as we mature, as most public companies do, we'll give guidance for the coming quarter, and I don't think it's prudent at this point, the numbers don't warrant it, but I think we'll see significant traction in the first quarter, we'll see nice growth throughout the year.

I would point you for a good model that's been developed in our last round that we did of, what was it, $21.5 million, it was done with Shaw and Partners. The research analyst there is very solid, he's written a very nice report, it has a financial model in it, it's got a revenue model. Thinking only of Brain Chip Studio and Brain Chip Accelerator, certainly 2018 is too early to start thinking about Akita revenue, but I think their model was something on the order of $7 million or so in 2018 revenue. I think it was a nice ground-up model, you know conversion of opportunities in the casino industry, conversion of opportunities in surveillance, so while those aren't our numbers, those are certainly numbers you can look to.

The company's opportunity to become profitable, you know this was another question, in 2018, that's really going to be dependent in some respects, how do you want to define profitability? Operating profit, meaning kind of steady state operating expenses out of your gross margin, I think there's a fine opportunity with the kind of model that Shaw and Partners has put together, and frankly we always like to overachieve, so I wouldn't look at them, look at their model as gospel necessarily, but then you have the Akita, which you could call it cap-ex, whether or not we can capitalize it or whether or not it'll flow through the income statement, but as you get towards the back end of the year when we launch, place, and route our physical design, when we go to buy the mast set, when we go to buy wafers, those will be incremental expenses above and beyond what our steady state operating expenses are.

Now with respect to steady state operating expenses, I guess we're currently running, just put a nominal number out, it's about $700,000 a month. The vast majority of that certainly goes to engineering resources. I do think it's important, I should have mentioned this when we were talking about our board of directors and management, you know part of the challenge of being a public company when you're in your formative years is it's pretty burdensome with respect to costs. We can spend in a year anywhere between $500,000 and $1.5 million on the expenses to be a public company. Now, we accept that, we want to be a public company and we want to create shareholder value, I've mentioned in prior calls that in 2016 there was something on the order of 140 M and A acquisitions of artificial intelligence companies, tens of billions of dollars in consideration paid, and not a single public market investor made a dime, because they were all venture capital backed.

The company decided to do the RTO, raise money in the public domain. There's costs that are associated with that, there's overhead that's associated with that, but it gives you as investors an opportunity to capitalize in a high growth market with a company that's really going to execute well. I just want to reinforce that, because this is an opportunity for public market investors that frankly doesn't exist anywhere else, but we have to put a whole team together, we have to put sales force together, we have to have a management team in the financial organization, compliance with ASX rules.

I'll go to the last slide just as a wrap up, little split on the screen as well. Last slide is a wrap up, the company is well capitalised. There was a question about how long the cash will last, you know at $700,000 a month, it's easy to do the math, and you also have to put in whatever your best judgment is on where you think we'll go with revenue, which our model has baked in, as I said, something nominally around $7 million, we'd love to overachieve, but we're well capitalized to get through the Akita development, build a big business, and call it vision, surveillance primarily, but also machine vision, Saffron, Rockwell Collins. We're well staffed in our incremental head count in sales, probably be more applications engineering, we'll have executives on the ground in all three continents that we choose to claim right now, we'll use manufacturer's reps in the United States.

They don't get paid unless they book an order, which is, you eat what you kill. Generally speaking, the engineering staff is well staffed as well, we probably need an engineer here, there, or elsewhere as we get into the Akita FPGA design or maybe further, you've got test and verification once you start to get silicone, so there's some disciplines that just don't exist in the company today. We have commercially available products for surveillance, again, you know software came in in July, hardware came out in September, been well received, dozens of software drops. People are now sending in RFQs for the hardware, as well as having delivered one to a major potential customer.

Nice established customer base, but again, it's very France-centric. Our goal is to really accelerate our penetration in the US, the balance of Europe, and leverage Greg and the team that we'll build in Australia. It's a very ripe environment for us, there's a whole lot going on in cyber, there's a whole lot going on in Fin Tech, there's a whole lot going on in surveillance in Australia. We have a presence, we get a lot of press, that's a great market for us to excel in. We have an increasing pipeline, then we'll give you I think some more real data about that when we talk in a month or so, but I'm very satisfied with the results that we've had, you know from the publicity that we got with Studio, the awards that we've won with Studio and Accelerator.

Revenue growth routes '18, I feel confident that we'll have revenue growth on a quarter over quarter over quarter basis. Profitability, I just touched on, and I think we went into some detail about the Akita development.

I hope that I answered all the questions, I hope I gave you a sense of the confidence that this management team has, both about the products that we have, as well as the really very exciting Akita development. We will talk again probably in another month or so after the 4C comes out, certainly you always have access to send follow up questions to that email address, and I'll try to be as responsive as I can, or someone on my behalf will be as responsive as we can.

Thank you again for joining, we'll talk to you soon.