 And we hear the IDTechEx show and hi! Hi, my name's Scott White, I'm Chief Executive of Pragmatic and what you've just been seeing is one of our wafers of flexible integrated circuits so these are very thin flexible integrated circuits much thinner than a human hair, just 10 microns or so It's a bunch of processors blowing in the wind It's very light In this case they're actually RFID circuits, radio frequency identification which is one of our key target markets So they get assembled onto an antenna like this You can see the antenna there and one of our flexible integrated circuits just crossing over the top of it there It's very small right? It is, although it's larger than an equivalent silicon chip and that actually has a benefit because we can use it to provide the crossover of the antenna So one of the advantages of our technology is not only is it thin flexible and very low cost but it also allows you to save costs elsewhere in the system by simplifying other aspects of the total solution So the RFID kind of antenna kind of market is pretty big right? It is, this year I think it's about 20 billion tags 20 billion, just this year only So this is how you get to trillions Yes, so that 20 billion is still limited by the cost and form factor of silicon based RFID which means it can only be used on relatively high value items So clothing in particular is the biggest market by far You can imagine putting it on an item of clothing that costs 100 dollars is quite justifiable for a tag that costs 10 cents say But if you want to extend that to everyday items that you might interact with like a bottle of water or a tube of toothpaste to be able to track that or have it engage with consumers then you need a much lower cost point and that's where our technology comes in By enabling that it can push the same kinds of applications to a much wider range of products So in my previous video there was a lot of comments about people asking why is it a wafer? Why is it in the shape of a wafer? So the main reason there is that a lot of the processing techniques we use are similar to what is used in conventional electronics manufacturing So actually when we're making it the plastic is coated onto a glass wafer so that it can be presented to equipment and look something like a silicon wafer So this can be used in standard wafer processing equipment so it can still get very good accuracy, very high yield and very mature equipment and processes in order to make this So using like a semiconductor industry is huge and making wafers the way they do it? It's not quite the way they do it It uses similar equipment, similar processes but we use different materials which are much lower cost and also the way we've adapted our process means the types of equipment we use as well as the processes are also much lower cost and much higher throughput And what is this showing? So this is effectively how it's presented to our customers So we produce it as a wafer on glass it then gets released from that but if you just have this very thin flexible wafer waving around it's very hard to actually do anything with that practically So it's presented to what's called a wafer frame so it's basically stuck onto here so then you can use very high speed pick and place or flip chip machines to attach this to the antennas It looks like it's more like it could be printed in a big thing like this instead of a wafer but that's the only way machines are built This is what the very high speed flip chip machines are used to handling So this is actually the easiest way to get products to market is to reuse the existing format that they're expecting to see What's a flip chip machine? Flip chip is basically a pick and place so it's a robotic system that can pick up each individual chip In the case of flip chip it picks it up, it flips it over so it puts the contacts face down and then presents it onto the antenna and bonds it to the antenna with a conductive adhesive How long does it take to make one wafer? So the production cycle time for our process is less than 24 hours So you compare that to a silicon fab where it would be typically a couple of months start to finish Start to finish? Yeah, it's less than 24 hours It takes a month? Months? In a silicon fab typically depending on exactly how you batch up the process you're looking at between 4 and 12 weeks typically That's a long time Yeah, so silicon you have some very high temperature processes that take quite a long time and also because of the way the equipment works in a silicon fab the process has to be batched so you don't have each wafer going straight from one process to the next In our case we have a single wafer flow so it's a very efficient movement of material through the system as well as each of the individual process steps are much faster we don't have any of the lengthy high temperature processes you find in a silicon fab And in the last video one of the highlights and we started the video talking about arms so that wasn't a specific size that was a little bit bigger than what you're doing now, right? Yeah, so we've seen a progression of our technology in terms of reducing the feature size and the overall density of circuitry it's still obviously not going to be at the same level as silicon and to be honest that's not what we're aiming for but if we can reduce over time the size of our circuitry then that means we can do more complex circuits in a reasonable footprint and a reasonable yield So this is the most recent version of the Cortex-M based SOC that we've been doing with ARM which is about half a square centimeter or so in size now and that's not actually in our latest version of the technology so we do think there's substantial scope to reduce that in size further This is similar to Cortex-M0? It's based on ARM's standard Cortex-M0 but it's an SOC so it includes the memory and IO and so forth as well it's not a full general purpose microprocessor so one of the things that intentionally we're focusing on with our process is how do you move away from completely general purpose architectures where you're using software and actually be able to optimize the hardware design at the point of tape out and we can do that because for us the cost and time scale for doing a tape out is far lower than in silicon So how far is it from something like this actually you can plug it in and start using? That's still probably a little way off Our main focus at the moment is scaling up production for our RFID circuits that's a market where there's, as I've described before, a very well established volume requirement and a clear need for our technology to be able to drive that into a wider range of products What we're doing on the Cortex-M project is really developing the roadmap for how do we add more functionality over time and that will expand the range of things that we can put our technology into But so for now it's going to be mostly for these kinds of things like this That's right So this is a fairly standard type of RFID tag architecture where it's on plastic substrate with an aluminium antenna This is an example of really where things are moving towards which is a paper substrate and a printed copper antenna This copper antenna is done by a partner of ours called Coprant and what that means is you have the ability to make it much more environmentally friendly both with the use of the paper substrate as well as the printing in order to produce the antenna Alright So what's next? What are you working on? What kind of challenges do you have? So as I said before, our main focus at the moment is on scaling up what we're doing on the RFID side The commercial take up for that has already been very strong Within the first two months after launching our first products we had orders for over 20 million chips and that's continued to grow since then That's also expanded into a wide range of different application sectors So while perhaps 12 months or so ago our main focus was on consumer goods packaging that's still a big focus but we've had other areas that have also become quite important for example healthcare and medical This is an example from a partnership we announced a couple of weeks ago with Shrine and MediFarm who are leader in smart labels and devices for medical and pharmaceuticals So which part is your part? Well you can't actually see it here because it's inside the label but basically it's an RFID enabled label so that actually you can identify uniquely the item as well as tracking things like used by date and so forth So you have all kinds of partnerships I saw your take at the smooth and sharp So this is about RFID stuff? Yes, so in the RFID market our direct customers are the RFID tag or inlay manufacturers so the companies that make these kinds of inlays which is basically a complete RFID tag with the antenna plus the chip So that's in companies like smooth and sharp that you saw as well as the likes of Avery Denison who's actually a strategic shareholder in the business They then sell to the label companies and so forth and the brands that are using the product at the end of the day but we also work closely with the brands ourselves in order to understand the use cases they have and what they want from our technology in the future So the previous way they would do this is by doing hybrid now it's going to be more kind of like a still hybrid but it's with the flexible CPU Yes, so the conventional way of doing tags looks very similar but instead of our flexible IC it has a bare die silicon IC and that would still come on a wafer frame like this but instead of our flexible wafer on there you'd have a silicon wafer that's been diced up into the RFID circuits So the model for our customers is very very similar but it just gives them a solution that has a much thinner, more flexible more durable form factor and much lower cost And so what's the status of the company? You're a big team, you have doing lots of cool stuff Maybe you can introduce Yes, so we have 85 people at the moment and at the moment we're at the transition really into volume manufacturing So the RFID circuit is taking off One of the things we've launched recently is what we call our FlexIC Foundry where we're working with third-party design houses to design their own circuits So perhaps to describe that I'll introduce Richard Price, our CTO Right, maybe you can swap the mic right here So this is getting a lot of views right now on my YouTube channel This stuff, this is really unique in the world, no? This tech Yeah for sure, as Scott just said we've recently launched our FlexIC Foundry service so that allows designers to use our process design kit and they can come up with new application designs and we can work with them then to go through multiple design iterations and fabricate those designs for them whereas typically in the silicon world that would be a six month process minimum that would cost you a few million dollars we're much much cheaper than that because of our different cost point and our much more rapid manufacturing So it allows you to do these multiple design iterations and really optimise the products and I think we said yesterday at the presentations we announced a couple of our beta partners there So ARM, as you know we've been working with for a number of years including on the Cortex-M development as well that we've worked with previously as part of a collaboration in a European project including developing an award winning design for RFID for playing cards along with Carter Monday and then some of the other partners there were talking things who were also producing RFID tags based on our connected products but they're moving into designing their own RFID proprietary circuits as well So I think it's a really big step for us to open this up and we're really excited about the types of applications that people are going to come up with and we look forward to making those as well So if you go over here on the wall that you have right there it says Flex IC Foundry So you are fab basically and you provide EDA tools kind of for fabulous semiconductor companies, designers to do their own stuff Exactly, so we've developed our PDK using standard EDA tools and that allows fabulous design houses to come up with their own designs As you said, some of the features there are we've got very low tape out costs much, much lower than they would get with a typical silicon ASIC and we can do these very rapid design cycles and we can work flexibly in a number of ways so we can either work one-on-one directly where a complete wafer will be designed for one partner or we can do multiple project wafers where we can manage that confidentiality across multiple partners So how low is the cost? Do you talk about this? It's like orders of magnitude Many orders of magnitude Yeah, so we're typically in the tens of thousands there So totally affordable to start doing crazy new things It really is affordable for anyone from universities to companies to start to think about new applications and crazy things But when you look at this chip over here, the arm, right? Hasun, what's the challenge to make it work? The challenge is really twofold One is we have a full production backlog really around RFID so we're delivering those products to our customers and the other is just continuing to evolve our process in terms of maturity so we can yield this type of product across the whole wafer and then it becomes really commercially viable We're going to be continuing to evolve our design rules over the coming years We know we can shrink the footprint of this much further and also improve some of the other performance metrics there as well by shrinking the critical dimension of our devices So in the previous video we did, you're talking about having a fab ready for billions of devices, right? How far are you in this process? How much are you shipping? So our first fab is called FlexLogic It's a completely automated self-contained clean environment around about 15 square metres of clean room This has a capacity of a billion circuits a year already We're designing our second generation which will have a higher capacity Based on the kind of connected RFID products that's going to be around about 5 billion circuits a year So we're pretty much there in getting into the billions The next step for us is moving into the tens of billions and then into trillions of units But you have the capacity, how about shipping that many? Are you getting there? Are you there? Yeah, so Scott said with the last sort of year to 18 months we've been transitioning from a technology business into a manufacturing So we're at that point now We've got orders in the tens of millions that we're starting to ship and next year is going to be about increasing that into even higher volumes So when we look at this armchip which is kind of like a Cortex M0 but it's not yet ready to work So what kind of architecture works now? What kind of architecture? So mostly what we've done today has been digital logic gates So that works very well in our technology We're adding additional analog functionality onto our road maps so things like analogs to digital and that's for a range of sensor applications So again with ARM we've been developing a machine learning processor as part of a project with Unilever This is for odor sensing and it's based around a flexible neural network and this is also a big exciting step for us and we think not only is that applicable for odor sensing but you can imagine food spoilage and other things where there's a volatile organic signature that can be detected in this way Alright so looking forward to what's next and this is happening a lot in the UK right? Is it Cambridge? It's a lot of our application and design work is happening in Cambridge We obviously work closely with ARM there but we're collaborating across Europe and North America and increasingly in Asia as well so it's not just a UK activity