Below I have just tried to summarise and bring together the main points from the Jan 3 webinar script. Cheers, Deena
Some KEY POINTS from 2018 BRAINCHIP Webinar, held 3 Jan 2019
# The edge is where high volume devices, whether it's IOT, medical devices, mobile phones or a wide variety of other devices exist
# There is no real AI solution at the edge today
# Think of Akidaas AI at the edge
# BrainChip’s goal is to be the most compelling solution to high volume markets
2018 Highlights
# Strong team; 35 people; 80% in research engineering; rest in marketing and admin
Engineering has focused on getting GPI done as well as keeping as many engineers as we could on the Akida development
# Very productive year for intellectual property
# In gaming. GPI (Gaming Partners International) a big plus. The ATS (Advanced Table System) should be production ready with the vision system in January
# In surveillance, we have a strong sales pipeline, although we haven’t turned that into revenue yet
# IC(Integrated Circuit) developmentis on schedule.
- Expect engineering samples in the 2ndhalf of 2019
- An extra quarter was added to hone the product definition as a result of customer feedback
# Akida has the ability to do CNN to SNN conversions. This gives us a quicker time to market
# Cyber Security. As a result of an acquisition of IP from Praxis University in Greece we have a fully functioning native SNN for cyber security, doing deep packet inspection, looking for malicious code that’s coming across IP traffic
# Although BrainChip Studio saleshave been weak the pipeline is robust. The expected sales cycle of 6 to nine months is more like 12 to 15 months
Sales optimised with internal advocate; then go through an evaluation, trial period, and then deal with the budgeting cycle (all takes time).
# Through SpikeNet we had customersin France; but now we’ve spread throughout Europe, Including the UK as well as the United States.
# Retooling the sales organisationwas a big part of what went on in 2018
BrainChip Studio Customers/Partners – Where are they now?
# Gaming Partners International(GPI). Customer interest was very robust from Macau and Las Vegas conferences. Enhancements as a result of specific requirements from customers and further enhancements are likely to be identified over the next couple of quarters.
GPI recently announced an agreement to be acquired by Angel. The Angel acquisitionis going to play out quite well for BrainChip.
Our team is working 24 hours a day with the GPI team to get this thing released as soon as possible in 2019.
# SN Tech. In September of 2018 we invoiced SM Tech for $609,135 and they disputed the invoice. We pursued it at a preliminary audit; however what we got from SN Tech was inadequate.
BrainChip is filing a Freedom of Information request through the Lockport School District before we determine any legal remedies.
# Quantum Corporationis a large storage supplier looking to integrate BrainChip Studio with what they call their StorNext Platform.
The integration, I'm saying nearing completion. We had to modify a doper which is kind of like a plug in so that you can just plug BrainChip Studio into their existing system
# Veritone– Integration nearing completion
# Unnamed largeVMS (Video Management System) provider. Contracts are in process. Once signed there will be a press release. This is one of the very large players.
# Safranhas continued to buy licenses but bits and pieces here and there. They had to put it out for tender offer. The tender offer was written around our specification. We've actually signed a purchasing agreement. We're probably looking at a couple hundred, maybe 300 channels to start at somewhere between 1000 and 2000 bucks a channel.
# Yvelinesis a restart. It looks like they're reassessing what analytics they want and how they want to architect the system. I'd take it out of our forecast but it is not dead.
Remember, all of these things that we're talking about are basically meeting 95 to 97 per cent gross margin
NDAs (Non-disclosure Agreements)
# With BrainChip Studioall NDAs have been at the request of the customer. As we move to full deployment, some will be quite comfortable with us talking publicly
# The NDAs that we request right now are associated with Akida, particularly prior to filing provisional patent
Let’s talk AKIDA – The heart and soul of the company
# With 1.2 million neuronsand 10 billion synapses, you can do a whole lot of processing
# With the Akida Development Environmentwe’ve got feedback and we’re working collaboratively to add some enhancements
# The data to spike converteris a very powerful invention or novelty. Data to spike can convert pixel data, lidar, radar, ultrasound, pressure, vibration, temperature, or sound to name a few. Can be used for any application where you’ve got large amounts of data
# The execution engineis unique in that it will have several training methodologies
# Akida will do unsupervised, autonomous learning.
# We’ve used interfaces and toolsthat are very well known.
# We’ll have a model zoodeveloped internally or collaboratively. This is basically a repository of developed applications.
# When the cyber thing is complete, we'll add cyber security
# The FPGA design is virtually complete. The FPGA design is for us internally, for us to debug and make sure that the logic works.
# The next step which will kick off inJanuary, called a hardened IC, is the logic design for the Akida device
# Manufacturing steps/processes. Lots of pieces to it.
- Logic design
- Synthesis and verification
- Physical design
- Cut the mask set
- Launch into fab
- Slice and dice it into chips
- Assembly
- Test
# Another path is to select a semiconductor fab company, and they do a cradled process from beginning to end and deliver a hardened IC. Cost for the IC is going to be a little higher, but you will have saved the capital cost of the expensive mass set. It really feels like a turn key provideris probably best for a first IC, but we haven't made that decision yet
# Akida as a Software Solution. Potential to actually generate revenuefor us before the IC comes out
Slide from webinar sums it up
# ADAS (Advanced driver assisted systems). They'll make no money on autonomous vehicles (AV) for many, many years. It's going to be a slow roll out.
For now it’s more about driver assisted systems. There are a number of paths to sales.
- Direct to major automobile manufacturers
- First tier vendors. We've talked about companies like Active and Valeo and Continental.
- Sensor manufacturers who provide integrated sub systemsfor the modules
# Medical devices. There is an increasing number of remote medical devices, whether it's heart monitors, blood pressure monitors, diabetes insulin pump, constant glucose monitors. These things are all becoming IOT devices
At the edge, there will be cyber security
# Akida at the edge, there's not a lot of competition because the spiking neuro networks are, we believe, the best way to reduce power
# Competition. They do not have enough horsepowerand we're far more cash efficient than many of the players that are talked about
Expand
(20min delay)