October 23, 2018 - 6:01pm EST by
2018 2019
Price: 17.42 EPS 2.483 2.818
Shares Out. (in M): 57 P/E 7 6.182
Market Cap (in $M): 999 P/FCF 14 14
Net Debt (in $M): 80 EBIT 179 206
TEV ($): 1,075 TEV/EBIT 6.13 5.29

Sign up for free guest access to view investment idea with a 45 days delay.

  • winner


Kemet and the art of horse rendering


Kemet has been a ‘frequent visitor’ to the pages of VIC, having been written up on FOUR previous occasions (in chronological order – and worth reading to give a background):

  1. 3rd April 2001 at $17.52 -

  2. 15th April 2003 at $7.50 -

  3. 2nd March 2010 at $1.35 -

  4. 30th August 2012 at $4.45 -

So what value do we bring in the 5th writeup?

  1. The market has under appreciated how Kemet has changed and the consequent moats

  2. We believe that the market is underappreciating the dynamics of the capacitor industry

  3. The market is dismissing the consequence shortages

  4. And is failing to appreciate the sustainability of the business

  5. The numbers look too low

In the following we will assume that the above write-ups have been read.

  1. The Changes at Kemet and the moats

In the fourth of the above writeups the partial acquisition of NEC Tokin was discussed. Kemet now owns Tokin completely (it exercised its options) and has integrated it. In retrospect because the deal took time to close (but the terms did not change) it was an extremely attractive deal (there was substantial cash within Tokin – the deal was first agreed in 2008 but only completed in 2017).

Previous writeups have discussed Kemet’s position in tantalum capacitors and its vertical integration (following the acquisition of Niotan = Kemet Blue Powder). We believe that the market has underappreciated the benefits that this has delivered Kemet.

In previous cycles the company had been dependent on an external supplier of tantalum ore and that relationship went sour leading to a collapse in Kemet’s gross margins (GM fell to 9% in 2008). The company has re-entered the Congo and invested in both production but also the social aspects (eg housing, schools for staff etc). Currently Kemet is one of the top two purchases of tantalum ore globally; with 50% of its supply from the Congo. The company is fully compliant with ‘conflict mineral’ requirements – this is becoming an increasingly important issue for many customers. Thus we believe that controlling its own supply chain for tantalum is an underappreciated moat for Kemet. (The writeup of 30th August 2012 provides more detail on this).

For reasons that will be seen later we would compare building a vertical supply chain for tantalum as attractive an industry as horse rendering – ie very few people are going to be attracted to it so the incumbent players are going to have a profitable niche.

The tantalum capacitor market is worth roughly $1.7bn worldwide – with Kemet’s market share at 33%; AVX is at 30%. Tantalum capacitors are smaller than ceramic capacitors and produce less (electronic) ’noise’ – however they are more expensive. Historically they were used for applications which were space constrained but demanded high quality – particularly Nokia handsets in its heyday. However, the increase in pricing of ceramic capacitors (discussed below) is leading to switching to tantalum.

The tantalum capacitor market is split into two parts – an older technology based on Mn02 (manganese dioxide) tantalum and a newer technology – polymer tantalum capacitors. The advantage of PTCs is that it is not subject to ignition risk and is even smaller than MnO2 tantalum (PTCs are, however, more expensive).

Kemet originally gained its technical expertise in PTC in the late ‘90s through a ‘technical sharing’ agreement with Tokin when it was part of NEC. This is the fastest growing part of the tantalum market (historically it grew at 10% per annum versus 3.3% for tantalum capacitors overall) and Kemet (including Tokin) has a 42% market share followed by Panasonic with 35%.  PTCs represent 45% of Kemet’s tantalum revenues.

We have touched on one moat with respect to the discussion on horse rendering – not many players want to be involved throughout the vertical value chain – yet for major OEMs ethical sourcing is increasingly important.

The second ‘moat’ is quality. It is worth reading the story of the ‘Capacitor Plague’ – see:

Consider would you, as a procurement officer, want to explain to Tim Cook, or the CEO of a major car company, that you saved 10 cents on a product but the result is a massive recall? (As well as quality, a related issue is availability. Imagine having to stop a production line due to the absence of a 10 cent capacitor?)

The third component of the moat (and related to the previous item) is knowhow and patents. We understand from Kemet that making polymer tantalum capacitors is non-trivial and there is significant associated intellectual property and knowhow.

As an aside (and for a bit of light relief) it is worth looking at the specifications of motherboards for gaming computers. For instance we just looked up the first board we found on the ROG (Republic of Gamers) website and found the second feature is:

High-Quality Solid Capacitors

High-quality capacitors withstand extreme temperatures of -55 to 105 a range that's 110% better than the industry standard

(Source: )

Similarly the following page on the Gigabyte website refers to the use of Japanese solid capacitors - and   (perhaps an echo of the consequences of the capacitor plague).

(Far be it for me to suggest some corporate mischief by suggesting you ask your head of IT whether the capacitors in your corporate servers are ‘high quality Japanese solid capacitors’ :-) ).

  1. The dynamics of the capacitor industry.

We believe that the supply / demand dynamics of the capacitor industry have changed significantly due to (i) withdrawal of capacity (ii) increased demand

Withdrawal of capacity:

Both Murata and TDK (Epcos) have announced that they are discontinuing manufacturing of a number of (ceramic) capacitors. It will be noted from the links below that some of the capacity withdrawals have already occurred over the last couple of years whilst some are due in April 2019 (Murata ie a further catalyst).

The key documents are here -

And here –

And here –

And here –

and here –

There are also documents pointed to here –

In a number of cases these capacitors do not have readily available replacements. These announcements have led to stockpiling of those specific sizes before they are discontinued (eg a number of Murata capacitors will cease production in March 2019). Furthermore, electronic and electrical designers are being actively encouraged to re-design their circuits to incorporate different (often tantalum) capacitors.

It would be a legitimate question to ask ‘well doesn’t the capacitor industry always go through a cycle of retiring some products’. My response to that is that it does, but reading industry publications suggests that the last couple of years and into 2019 there are more lines being retired than historically leading to shortages (see section on evidence below).

Another related question is why are the large Japanese players withdrawing capacity. We think a better way to express the situation is that they have got fed up losing money and are now deploying capital (and capacity) towards those devices with more profits (ie moving from larger capacitors to smaller capacitors which have higher margins).

Increased demand:

One might think that with stockpiling there would be no issue. However demand has been greater than anticipated throughout the electronics industry due to (i) increasing use of capacitors in handsets – for example one of Kemet’s IR slides shows that the iphone 6s had approximately 500 capacitors whilst the iPhone X had over 1000 (ii) the development of electric cars and electrification of even internal combustion engine (ICE) cars (for instance a traditional ICE car might have 2,500 capacitors; the Tesla Model and X both have over 10,000 capacitors whilst the Model 3 has 9,200) (iii) generalised electronics demand (particularly computers / datacentres including the whole associated power infrastructure).

A number of circuits in ICE, PHEV and EV cars require more expensive capacitors – in some specific cases the value of a single capacitor can be over $100. (Note that historically capacitors were often priced in at a cent or less in volume). Though it is hard to be specific it does look there may hence be a small improvement in ‘mix’ (separate from the move from MLCC to tantalum / polymer which also is a positive mix shift).


I am sure the ‘tough audience’ of VIC will be sceptical of me suggesting that there is a supply / demand mis-balance in the capacitor market. So perhaps the following ‘open letter’ from TTI (a major electronics distributor) might be of interest:

April 11, 2018

Dear Valued Customer,

As you are likely aware, there is a worldwide shortage of commonly used multi-layer ceramic chip capacitors (MLCC). As a global distributor of the leading capacitor brands, TTI is close to the situation and we offer the following perspective, assurance where possible and suggestions for the future.

There are a number of contributors to the shortage and we believe once you understand the scenario you will likely agree with our team of specialists that the conditions of this market are different than historical shortages and allocations of components – the most critical difference being that this time of constraint will last longer.

The underlying cause is economic and largely pertains to case size 0402 and up, both low and high CVs, though case size 0201 in some values are included.

Large case size components have been subject to cost-down pressure for so long that meaningful profit for the manufacturer has been squeezed out. Statistical evidence proves that since the market recovery after the 2007-08 recession the production of MLCCs doubled but the value of the market remained flat, that is until last year when market value increased only slightly over units. We have reached a point that some manufacturers are concentrating their future activity on smaller case sizes and less commoditized technologies of the market like high voltage and flexible termination, with a particular focus on high-reliability end-markets such as automotive and medical. So while manufacturers are adding capacity, it’s generally not for commercial commodity parts – making this market condition even more difficult to navigate.

This is not to say that there is no incremental capacity in the works. As pricing reacts to increased demand some capacity is being added for commercial parts but the turnaround time is longer than normal. One holdup is the availability of new equipment that comes from a small group of equipment makers whose backlogs have also grown with the economy.

In early 2017 our product specialists began preparing for this market by aggressively expanding our inventory position on MLCCs with the goal of shielding our customers from expanding lead times and price increases. To further ensure we could support existing customers, we decided to decline orders from new customers for constrained and allocated parts (this situation extends to other dielectrics, resistors, inductors and power semis) – our team intuited that this situation could extend well into 2018.

Now that we are in 2018 conditions are even worse than we anticipated – our product allocation from suppliers has been based on previous year’s consumption so even with our decision to reserve our inventory for existing customers, we will not have enough to cover the increases you are experiencing in your businesses. It is not unusual right now to see true demand increases (not artificial as a result of hedging) from customers of 30% or more year on year. The supply chain is further complicated by the fact that there are a number of buy-resell agreements among MLCC manufacturers and we are already seeing unexpected shortages as a result. As these come into play there may be circumstances where we will be forced to advise our customers that the product is simply not available and the only recourse is to identify and approve an alternate component.

So what are we doing? 

At TTI we are actively determining when we will lose inventory coverage, customer by customer, part by part, so that we can give each of our existing customers as much notice as possible. At the same time, we are presenting each customer options in the form of alternate suppliers, alternate part numbers that are perhaps higher performing but still drop-in replacements, and even alternate technologies. During the tantalum shortages in the early 2000s, many customers switched to MLCCs. Reversing that could provide near-term relief, especially as it pertains to polymer capacitors as a backup for some high CV ceramics.

The most important thing you can do is evaluate alternate part types and suppliers for both low and high CV MLCCs. 

We understand the time and cost related to this effort, especially on existing products, but we urge you to begin as it may be your only path to ensuring your production lines continue to run. 

The intent of this letter is not to cause panic but rather to increase awareness. TTI is as well or better positioned as any in the industry to support you through these shortages and as a supply chain customer of TTI, you will always be our priority. Our intent is to minimize the impact that any of these parts that we supply may have on you and we believe that starts with information and communication.

Source: (our highlights)

As will be seen from the above open letter, despite ‘stockpiling’ TTI has found that conditions are much worse than it expected.

If this is not sufficient we recommend viewing the following video featuring two elderly gentlemen from Omaha who are the indirect owners of TTI - . The key point these gents make is that ‘electronic components have absolutely taken off’. However, if nothing else, the commentary on horse rendering makes it a worthwhile watch. (Charlie uses horse rendering as an analogy to component distribution; I would suggest that vertical integration along the tantalum polymer value chain from sourcing ore to manufacturing is another ‘horse rendering’ industry).

Other distributors have also made comments on capacitor shortages eg Advanced MP - see ; Lantek - In fact we would recommend googling ‘capacitor shortage’ and ‘MLCC shortage’ (MLCC = multilayer ceramic capacitor). The general consensus appears to be that shortages will last at least to 2020.

Part of our hypothesis that shortages of MLCC is leading to active switching to polymer tantalum capacitors. An example of such an exhortation is provided by Arrow Electronics (another major component distributor) – see (Note that the preceding website also explains when it is appropriate to switch from MLCC to PTC and when not). Avnet is also suggesting an MLCC to PTC switch – see

We should also highlight that adding capacity to the capacitor market requires equipment which takes (our estimates from conversations with Kemet) 9 months plus for kilns etc – assuming you have the technical knowhow. And the withdrawal of the Japanese from certain sizes illustrates (as highlighted in the TTI open letter) that the industry is no longer prepared to take losses.

On the demand side, it will be noted that the TTI letter refers to underlying of 30% year on year – we believe that the market has missed this.

In this context, though only indirectly related, we would also highlight Intel’s recent open letter where it reported that cloud revenues had grown 43% in H1 and that even the PC market was growing (see ).

Perhaps it’s ironic – or cause and effect but the withdrawal of supplies comes after the DOJ broke up a price fixing cartel for capacitors – see ,

(Note that this did involve NEC Tokin; Kemet was also identified – see and  ).

C. The Consequences of Shortage

As a rule of thumb ‘passive’ components represent about 90% of the components on a circuit board but only 10% of the value; and vice versa for integrated circuits (ie chips – the ‘active’ components). There are lots of different types of passives from resistors, chokes,  inductors, transformers etc. Hence capacitors are only a fraction of the entire ‘bill of materials’.

Depending on the application a passive component may be only a fraction of a penny – but the finished product may be worth hundreds, if not (tens of) thousands of dollar. Therefore customers are willing to pay for components they need or to gain availability. Depending on the product lead times are ranging from 8 – 52+ weeks or are on allocation (ie if you are not an important customer you get zero) – see   (indeed we believe that some manufacturers and distributors are turning away new customers).

In addition a concentration of suppliers, a tight market and thousands of customers is altering the supply / demand relationship. [Investor Relation slides from Kemet suggest that 50% of ‘legacy’ Kemet’s revenue came from 1,000 customers and 50% from the remaining 182,000 customers; for ‘legacy’ Tokin 99% of the revenue came from 630 customers – see presentation January 2018 at the Needham Growth Conference]. (Note that Kemet actually sells to OEMs (45%), distributors (42%) who then sell on to smaller customers, and EMS (13%). [EMS = electronic manufacturing services ie companies such as Jabil).

Various industry sites suggest that prices have risen and will continue to rise. During 2018 for instance some reports suggested substantial price rises - see and  We think that Kemet has invested in longer term customer relationships and so has not increased its prices as aggressively as the spot market – but that means it has ‘pricing headroom’ in the next year.

For major companies on contract however we suspect the price rises have been lower but there have been volume commitments over a number of years. For major customers the issue is not just price but availability; it appears that conversations between Kemet and customers have been focussed on availability.

On 7 Sep 2018 Kemet filed a 8-K reporting that an undisclosed customer had advanced it $36M (as an interest free loan) to secure new capacity (see ). We think the market has missed the significance of this announcement. (VIC readers may wish to guess who the customer is – we have suspicions.) We think this release also reflects the inherent conservatism of the company (and indeed the industry) – ie no public press release, no headlong investment without customers funding it or strong visibility.

It should be acknowledged that both Kemet and others have talked about adding capacity – however, given the growth of demand the industry consensus is that shortage is likely to last until 2020 at the very least. (We estimate for instance that Kemet will add 10% to its capacity over the next two years).

D. The sustainability of the business

Once a component is designed into an electronic circuit it is usually not worthwhile for a manufacturer to change the design (except when components are not available as with MLCCs currently). (Thus the industry and the high end leaders (eg Kemet) have more pricing power than is generally realised.)

Hence in many ways the company has an ‘annuity’ stream. However that clearly depends on the volumes of end products sold.

The leadership position in tantalum and, in particular in tantalum polymer provides growth and exposure to a secular trend to use higher quality products. We believe that it is reasonable to assume that short cycle (ie 1 – 2 year) products where quality (particularly of sound) and size are important are more likely to move increasingly to tantalum polymer ie handsets.

There are also some attempts by some electronic designers to reduce the capacitor count during redesign. Often such redesigns can require better quality / more efficient capacitors – thus tantalum becomes more attractive.

It is also worth considering the voltages used in electronic circuits. Typically desktop PCs would have ‘rails’ running at 12V, 5V and 3.3V (see ). In comparison handsets often run at 5V to 1.2V. But cars (ICE) are moving to 48V. The Tesla uses cells generating 1.2V (if I recall correctly) but then builds the voltage upto 350 – 400 V (see ) . As a general rule of thumb a higher voltage requires a bigger capacitor which generally means a more expensive capacitor.

Hence it seems reasonable to assume that the increasing electrification of ICE cars and the move to electric cars will see not only increasing numbers of capacitors but also a positive mix shift.

E. What does this mean to numbers

Kemet is already ‘cheap’ versus its US listed peers. Furthermore recognising that Kemet is fully independent makes it arguably even more interesting. (Vishay is partially owned (72.2%) by Kyocera ie Panasonic’s parent).


















Looking forward it seems to us that distributors are anticipating price rises of 20 – 30% in H2 2018 – H1 2019. In addition to this, as discussed above, Kemet should benefit from the secular increase in electronic devices and a move towards tantalum and tantalum polymer.

For the following it should be noted that the Kemet fiscal year is to the end of March.

If we assume prices increase for Kemet by only 5% (assuming it exchanges long term supply agreements for less price increases), and 5% gain from each of the increase in volumes and the mix shift we easily get to a 15% increase in revenue in 2019. In comparison the current Bloomberg consensus shows $1.348BN for 2019 which represents a 12.4% increase in revenue yoy.

(Note that some of Kemet’s products are subject to tariffs eg from Mexico to the US. Currently we believe that Kemet is adding these tariffs on top of the contracted price but this may also cause confusion on how much prices have increased. We believe that Kemet will be able to increase prices by 20% + but in the preceding paragraph we intended to show that even a 5% increase allows Kemet to reach its numbers. Note also that the price increases are back end loaded in the year.)

For FY 2020 (ie to March 2020 – so 9 months of 2019) Bloomberg is showing revenues of $1.435BN ie an increase of 6.4%. Frankly with even conservative numbers we get double that growth (there will be a full year of price increases). We appreciate that the industry has typically been a price taker so volume increases have been offset by price cuts. However if our hypothesis is correct (and remember that some of the Murata products only go end of life in April 2019) then even flat pricing means that the consensus growth estimates are wrong.

Historically the company tried to maintain its margins (in an environment where it was a price taker) by continuously cutting costs. Thus in an environment with price stability (or even price increases as we believe), the margin expansion should continue to be significant as the cost cutting programmes are also continuing.


We hope we have shown why we think Kemet is interesting. The withdrawal of capacity from some manufacturers has led to a reduction in supply at the same time as there is a dramatic increase in demand. The secular move to higher end capacitors also provides some insulation from any market downturn so the company also benefits from a mix shift.

Even without any of the above drivers the company is valued at a discount to its peers; its position in polymer tantalum capacitors should argue give it a premium.


I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise do not hold a material investment in the issuer's securities.


1. Further withdrawals of supply by Murata in the capacitor market in April 2019

2. Price increases by all major players between H2 2018 and H1 2019

    sort by    
      Back to top