Thin Film Electronics — Update 8 April 2016

Thinfilm is the global leader in the development and commercialisation of printed electronics. The company prints electronic components and integrated systems and is expanding its production facilities to achieve high production volumes at low unit costs. It has developed and patented a range of printed, rewriteable, non-volatile memory labels using its own as well as non-proprietary technology such as near field communication (NFC). We see broad applications across most sectors for NFC OpenSense labels, but for its authentication applications the company is targeting the beverage, pharma and high-value consumer goods industry. It is targeting the retail sector with its electronic article surveillance (EAS) tags, the food and perishable drugs industry with its sensor labels and branded goods with its brand protection labels (to be produced under licence by Xerox).

Over the last six months, Thinfilm has doubled the number of high-potential partnerships and/or pilot orders for its products, in particular NFC OpenSense. We expect these partnerships to translate into major volume order inflows from the second half of this year, resulting in a strong uplift to revenue in 2017 and 2018 and a positive impact on investor sentiment. The expected move to R2R production in 2017 should cater for this growth and result in reduced unit production costs. We see the key concerns for investors as the risks surrounding transformation to R2R production and the likelihood of further equity financing being required in 2017 to fund the final quarters of cash burn and capacity expansion before the company becomes self-financing.

We have reduced our 2016 revenue forecast by 8%, primarily based on our assumption of sales of hand-held ASICs, which read memory labels to be produced by Xerox, but with little impact on earnings as they are sourced externally and resold at very thin margins. Our forecasts for 2017 and 2018 are for improving revenues and margins, with the company achieving a net profit in 2018.

Our DCF model post the Woodford equity issue gives rise to a valuation of NOK7.52 per share, up from the previous NOK6.69, mainly reflecting growth in our earnings expectations. This assumes that the company takes the decision this year to skip the proposed expansion of sheet-based capacity from 40m to 140m units and moves instead straight to R2R production in 2017, with $9m capex this year vs $20m with the intermediate step. Without this, the valuation would be NOK7.34.

Thinfilm designs, develops, produces and licenses printed electronic systems. It is the only pure-play listed printed electronics company and is leading the way in terms of building the printed electronics ecosystem that is a key enabler of the much-touted ubiquitous computing and IoT. With 270 patents and patents pending in printed electronics and NFC, the company has received multi-million unit orders for its EAS and OpenSense products and is now looking to scale that into the tens and hundreds of millions.

Thinfilm’s vision is to lead the creation of the new semiconductor industry, which uses printing processes rather than lithography to bring low-cost electronics to ultra-high volume applications that traditional electronics cannot reach. The industry appears to be in a similar position now to that of the semiconductor industry in the mid-20th century and we see the potential for it to follow a similar growth trajectory over the coming decades. The company has announced its ability to add new 1bn OpenSense-equivalent unit production lines at its San Jose plant for investment of c $25m – substantially below the $bn+ investments required by the semiconductor industry for similar capacity. We expect clarity on the timing and scale of transition to R2R production later this year.

Thinfilm started by developing printed memory products and is now transitioning from a development company into a product company. It is optimising its supply chain and manufacturing facilities to be able to cope with the demands of large-scale FMCG customers. The complexity (and therefore value) of products is expected to grow over time. Thinfilm intends to develop the technology, demonstrate its commercial viability using internal production and then out-license production to third parties. It has signed a licensing agreement with Xerox for outsourcing Memory production and Xerox is adapting one of its facilities to provide 1bn+ memory label capacity. It is intended that this model will be applied to each generation of products.

The start of the production process for any label is the printing of components and circuitry onto the ‘substrate’, which is typically a clear plastic sheet or metal foil in the case of the PDPS (printed dopant polysilicate) die. Not all components of the system are printed at the same time, as this reduces the yield of the complete system. Some specialist components such as batteries - although from 2016 the company will start to print its own based on Imprint technology - and sensors are also acquired from third parties and therefore need to be assembled into the system rather than printed as a complete system. This stage of printing components on a substrate is known as the front end. Once all the components have been printed or acquired from third parties they then need to be assembled into a complete system. This is done using automated processes similar to those used for traditional electronics assembly and is subcontracted to Thinfilm’s partners in South-East Asia. The amount of back-end assembly required varies depending on the product.

Thinfilm Memory is a standalone product and therefore does not require any integration, but the more advanced sensor system has multiple components and back-end assembly can therefore account for almost half the cost of manufacture. However, this should come down as volume is scaled and processes are optimised. Indeed, for smart labels the key bottleneck at present is on the back end, which is the assembly of the components into the final product. De-bottlenecking should be finalised around the middle of this year, enabling the focus switch to building the order book.

Thinfilm’s management has extensive experience in the printed electronics industry and the technology industry generally. Davor Sutija, CEO, has a PhD in chemical engineering and served as senior vice president, product marketing at FAST, which acquired Opticom, the original owner of Thin Film Electronics in 2006. Dr Christer Karlsson, chief technology officer, was previously deputy research director at the Swedish National Defence Research Establishment. Dr Peter Fischer was previously CTO at Plastic Logic and is on the board of the OE-A. Dr Henrik Sjoberg, SVP of product management, previously worked at ACREO, a current partner of Thinfilm, and John Afzelius-Jenevall, CFO, joined Thinfilm in August 2013, before which time he was VP of corporate development at Norwegian conglomerate Orkla ASA and portfolio manager Nordea IM. He is a CFA charter holder with a Master’s in engineering and Bachelor’s in economics.

NFC OpenSense is Thinfilm’s flagship product consisting of a near field communication (NFC) circuit and a sensor capable of detecting if a container has been opened or refilled. The former allows users with NFC-enabled smartphones (currently estimated at 1.2bn globally) to tap the tag and access content (information, promotional offers and usage tips) to improve their experience. The seal technology is a powerful anti-counterfeiting tool, with two-stage authentication providing robust security. Having fast read capabilities (100x faster than encrypted tags), NFC OpenSense labels are also suitable for high-volume production lines. The expected volume price point is 30-50c.

Allied Market Research forecasts that the global anti-counterfeiting packaging market in food and pharmaceuticals alone generated revenue of $57bn in 2014 and is forecast to grow at a 14% CAGR to $143bn by 2020. Brown-Forman, owner of the Jack Daniel’s brand, was reported in the FT in 2012 as stating that one-third of the world’s alcohol consumption is “illicit”, which includes unregistered and counterfeit alcohol. China has a particular problem, with specialist Nick Bartman estimating that 70% of wines on sale in 2013 were counterfeit. The IWSR estimated consumption of Chinese spirits at 14.4bn bottles in 2012 and wine consumption at 1.8bn bottles in 2014.

The tobacco sector is another high-potential significant market. In markets such as Vietnam, the government estimates that 25% of the 4bn cigarette packets sold in Vietnam are untaxed or illicit, while in the UK c 10% are thought to be illicit and illicit alcohol is estimated to cost £1.2bn in lost tax revenues. Thinfilm recently announced a partnership with a regional company SmartSign to work with the Vietnam government to combat this fraud. The potential of Thinfilm’s technology has also led the World Customs Organization to form a partnership with Thinfilm, which could result in adoption of the NFC OpenSense solution across a number of global markets.

With the wine and spirits market subject to a high degree of marketing for differentiation, as well as counterfeiting, the first interest in NFC OpenSense came in early 2015 from Chinese-owned Australian winemaker Ferngrove and global drinks company Diageo. We expect that these companies are pilot trialling the product in the Chinese market, a global counterfeiting blackspot.

More recently, with more than 1.2 billion people estimated to have NFC-enabled phones, Thinfilm’s NFC labels are also attracting interest solely for their consumer engagement advantages. Medical manufacturer Ypsomed is currently incorporating NFC OpenSense into its autoinjectables. The labels will enable the patient to check if they have administered the drug correctly and inform their doctor via the NFC chip that they are up to date with their medications. At the same time, a leading (unnamed), fast-moving consumer goods company and two other leading packaging companies (Constantia and Jones Packaging) are trialling Thinfilm to create smart packaging. Arc (shopper marketing and part of the third largest global ad agency) is looking to offer the tags to its clients as CRM tools – these include McDonald’s, P&G, Kellogg’s, Coca-Cola and Miller-Coors, and Tata Consulting will offer Thinfilm tags as part of its shelf label solutions.

Anti-counterfeiting technology and products come in a very broad range of form and function. Holograms are still widely used, while at the top end of the range RFID is one of the fastest growing technologies despite having much higher prices. The major drawback of these technologies, in comparison with NFC solutions, is that they do not provide a reliable means for consumers to authenticate an item before purchase, and therefore do not allow them to shop with confidence.

The deployment of NFC in this area is still in its infancy. Nevertheless, French company Inside Secure is working on a similar product to Thinfilm, trademarked CapSeal, which Remy Martin is currently field testing. The device is located in the cap of the bottle and contains an NFC chip. We believe that Thinfilm’s use of lower-cost printing methods should give it an edge over the French product in terms of cost and enable faster scaling as demand increases. At present, we see the added competition as a positive in helping to develop the still immature market for the technology.

EAS tags are theft-prevention devices, which attach to retail products and cause an alarm to trigger if the item is taken out of the store without the tag being disarmed or removed. There are many types of systems, but Thinfilm’s product is compatible with the global base of 8.2MHz RF EAS infrastructure, which supports more than 5bn existing EAS labels in stores.

Market leader, Checkpoint, estimates the addressable market for retail loss prevention at $1.1bn, with a current CAGR 2-3% pa driven by stock shrinkage due to theft of c 1.5% globally. The top two players in the EAS market are Checkpoint Systems and Tyco Retail Solutions (branded Sensormatic), with the next five being Nedap, which Thinfilm is already supplying with its tags, All-Tag, Sentry Technology Corp and TAG Co. Checkpoint sells both Radio Frequency (RF) and Acoustic-Magnetic (AM) tags. Tyco Retail claims its EAS solution is used by 80% of the top 200 retailers and tags 4-5bn items at source each year. Sensormatic primarily uses AM EAS technology, which it claims outsold RF technologies by a ratio of 3:1 between 2000 and 2008 due to its smaller size, wider detection radius and lower false alarms.

Thinfilm’s key advantages are that once de-activated in the store, its tags stay permanently deactivated, unlike RF tags, which can come back to life if washed or flexed (the “Lazarus Effect”). They are also thinner and more flexible than bulky RF and AM clothing tags, and can be incorporated undetectably into garments and shoes to give a better shopper experience. As they can also be applied to garments and shoes during production, they incur lower labour costs compared with RF tags, which require shop staff to apply and remove them and have the potential to damage garments. With major operators such as Checkpoint often installing store infrastructure at low margins with the strategy of charging high prices for the tags and maintenance, compatibility with existing systems is a key feature of Thinfilm’s labels.

Thinfilm tags have a wholesale pricing point of around 5c, which compares with prices for multiple-use Sensormatic clothing tags of 25-60c for orders of 1,000 units and over. Since October Thinfilm’s labels have been tested in a field trial in shoes in the retail outlets of one of Nedap’s European clients. At the customer’s request, Thinfilm is now applying an adhesive layer to its tags and is seeking to have its labels certified for use in womenswear and denim apparel. The company is close to completing a 13m unit order from 2015 and is currently negotiating follow-on orders with Nedap, as well as the expansion of the use of its products from shoes to women’s apparel and denim, which could lead to orders in multiples of 100m units.

Thifilm’s memory labels use precision-manufactured printed ferroelectric polymer to create a unique analogue signature that is practically impossible to forge. The labels can store data such as product IDs and can only be read by a hand-held device containing an ASIC that can automatically verify the memory based on the analogue signature being authentic. The product’s attractions are that it is highly secure, as well as low cost, with a pricing point of approximately 5c per label.

Thinfilm began sales of these labels for low-cost testing of the authentication of product refills such as printer ink or automatic air fresheners. However, in 2015 it licensed production to Xerox for use in brand protection labels, ie labels ensuring that counterfeit products are not passed off as branded products, and for government/custom agencies use as tax stamps.

Xerox is a global leader in business services, digital printing and document management and unveiled the two products it intends to sell featuring Thinfilm’s Memory in September, 2015. THIN materially completed the transfer of the Thinfilm Memory technology and made deliveries of R2R test equipment to Xerox in Q415m as part of the latter’s construction of a 1.3bn unit capacity plant in New York. Thinfilm should generate licence fees from the sale of its technology in the low double-digit percentages. Based on the abovementioned price point assumption, we have assumed that the company generates around 0.6c per unit from initial sales, falling to closer to 0.4c as sales volumes grow and prices fall.

The global security printing market was estimated in 2014 by Smithers Pira, a consultant on packaging, paper and printing, to grow by 5.9% pa over the five years to 2018, to $35.3bn. We see strong potential forecast upside for Thinfilm’s products given this large market. We estimate that our current forecasts for Xerox sales constitute only a 0.2-3% market share in 2019.

Thinfilm is developing a range of printed Smart Sensor Labels with a variety of functionality (eg temperature, humidity, light), as well as NFC connectivity for use in food, temperature-sensitive pharmaceuticals, wines, flowers and temperature- and time-sensitive chemicals.

Temperature sensors are currently the largest of the potential addressable markets for Thinfilm’s sensor labels. Estimates of the size and growth of this market vary widely. Gartner estimates it is currently worth $1.4bn pa, while MarketsandMarkets estimates it to have been $4.8bn in 2015 and set to grow at a CAGR of 5.1% to $6.1bn by 2020. Techavio forecasts an even higher CAGR of 16% during 2014-19.

We expect increased sales of fresh food and temperature-sensitive medicines to be a major driver of demand, as well as increasing sophistication of sensor products and the development of the IoT.

Currently, the major products on offer are low-cost chemical labels, which often require the user to interpret a colour change to determine if key temperatures have been breached, and electronic sensors with a much higher degree of accuracy and sophistication, which are at the higher end of the price spectrum. Thinfilm labels fit between the two in terms of quality/features. With clear readouts and NFC functionality, they enable rapid checks of store stock for safety. They are thin and flexible, with the option for dual temperature measurements (eg to prevent spoilage by freezing and high temperatures), but they are not yet as high spec as the top-end, silicon-based products.

Thinfilm’s pricing strategy should attract market share from both ends of the market by positioning its products between the two product types. A key advantage going forward is the scalability of its product versus the chip-based products, particularly with the move to R2R production, which should enable the company to become even more price competitive.

Thinfilm has garnered substantial interest from major players in the sensor market including Temptime, a US-based private company that manufactures chemical-based, time- and temperature-sensitive labels. It has also entered into a commercial distribution agreement with PakSense for the distribution of Thinfilm Smart Labels. PakSense has over 1,500 customers in 70 countries and produces traditional silicon-based sensor tags to monitor the storage condition of goods through the distribution chain. It has also been in partnership with major packaging supplier Bemis since 2012, which gives Bemis access to Thinfilm’s brand-protection technologies, as well as the intelligent packaging solutions. Bemis produces c 200bn packages a year and generates revenues of c $5bn, so capturing even a small percentage of these package sales would generate significant revenues.

Exhibit 3 shows a summary of our assumptions out to FY18. Exhibit 4 gives a further breakdown of our unit sales forecasts by product. In unit sales terms, we see the strongest growth in the short term on sales of brand protection labels, which have been licensed to Xerox. Xerox has earmarked annual capacity of over 1bn units for its production of these tags and, with the help of its extensive distribution capabilities, we assume that sales reach 1bn units in 2017 and 1.5bn in 2018. Based on our price estimates for 2018 of 3.2c/unit, this represents a c 0.14% share of the security printing market based on forecasts by Smithers Pira of $26.5bn in 2013 and a 5.9% CAGR to 2018.

We forecast NFC OpenSense unit sales to grow to 153m in 2017 and 506m in 2018. This reflects the significant pipeline of potential order flow and the substantial market for anti-counterfeit and CRM-enabling products. Our forecast, incorporating our product price estimate for 2018 of 32.8c/unit, represents c 0.15% of purely the global pharma and food anti-counterfeiting market, based on estimates by Allied Market Research of a $143bn market by 2020. Unit sales of EAS labels, which reached 11m in 2015, are forecast to grow to 151m in 2018 with product prices at 5.9c/unit. This represents a 0.8% market share based on Checkpoint’s market share estimate of $1.1bn in 2015 and its forecasts of market growth of 2.9% through to 2016. Due to the limited visibility ahead of the product launch in H216, we forecast sensor and smart unit sales of only 1m and 6m this year and next. We see significant scope to increase our forecasts after this point.

We assume the main costs included in COGS are ink costs, back-end assembly and sublicensing fees for third-party components. At low volumes, ink costs and back-end assembly costs are relatively high per unit, but this should fall as volume increases. Moreover, margins should widen significantly as production moves to the R2R plant and greater economies of scale are realised.

Thinfilm will complete the expansion in capacity of its sheet-based line from 28m to 40m+ OpenSense-equivalent units in Q216, which required $4.2m investment. Management is currently assessing whether to continue to expand the sheet-based system to 120m units and then add a 1bn+ unit R2R line or to switch immediately to the construction of the R2R facility, which is expected to cost $20-25m.

At an estimated cost of $12m (based on spending thus far completed towards management’s estimate of a total $17m cost to reach 120m units), further expansion of the sheet-based plant would require a much larger 15c capex per unit of annual capacity, compared with 2.0-2.5c per unit with the R2R plant. The R2R plant should also allow the company to realise greater employee cost savings. As such, we have assumed that management will decide to skip the intermediate expansion phase and assume capex this year of $1m for general plant improvements and $8m for the 30% down-payment on the R2R plant. For 2017 we assume that capex will be $17m, representing the remaining payment for the R2R plant. With an estimated $21m cash burn from operations, we assume that a further $40m funding will be required in 2017 (which we have modelled as short-term debt), after which the company should become fully self-financing. We assume that another 1bn+ R2R line is installed in 2019, but from 2020 we assume that all further sales of its products are a result of licensing agreements, leading to lower capex, opex and revenues.

Change in earnings forecast

Thinfilm reported a 9.7% higher than expected revenue number in 2015, reflecting the addition of substantial, largely one-off tech transfer fees from Xerox in Q415. These were offset by high material costs arising from heavy testing of the production line leading to a broadly in-line profit.

We have revised our forecasts, increasing our NFC OpenSense revenue expectations by $0.2m. We have nevertheless cut our revenue forecast for 2016 by 8%, primarily on reduced expectation of sales of hand-held readers for Memory labels by $1.1m (as we expect the large order sizes at Xerox to translate into a lower than previously forecast ASIC-to-label ratio) as well as a pushing out by one quarter our expectations of the inflow of follow-up EAS orders from Nedap (down $0.6m), which were originally anticipated in late 2015, and the start of sales at Xerox, which we estimated for Q116. As ASIC are produced externally and sold at very thin margins, this revenue cut has very little impact on earnings. On the expense side, we have restated our forecasts to include COGS in operating costs to bring them into line with reported earnings, and increased material costs overall to reflect guidance of expected ongoing testing of production capacity this year.

With Thinfilm at an early stage in its development and with no direct peers, we believe DCF is the most appropriate valuation method for the company. We have assumed a 2016-20 revenue CAGR of 173% as production is ramped up, but then a 0.5% decline to 2026 as the bulk of new sales are derived from licensing. We forecast the EBITDA margin to rise to 25.6% in 2020, but to experience some weakness in 2021-24 on assumed product price and licence fee erosion. We have reduced the notional tax rate in our model from 33% to 25% to reflect the current Norwegian tax rate. This positive impact on the valuation was partly offset by the strength of the NOK and dilution from the Woodford share issue. We have used a WACC of 15% and a 3% terminal growth rate in our model.