|Shares Out. (in M):||126||P/E||0.0x||0.0x|
|Market Cap (in $M):||4,765||P/FCF||0.0x||0.0x|
|Net Debt (in $M):||1,491||EBIT||312||267|
|TEV (in $M):||6,256||TEV/EBIT||20.1x||23.4x|
Sign up for free guest access to view investment idea with a 45 days delay.
SHORT TDK Corp. (6762 JP)
TDK Corp. is a Japanese manufacturer of electronic components with a market cap of US$4.8 billon. The Company generates >100% of consolidated EBIT from the production of heads for hard disk drives (“HDDs”). Analysts consider this segment the crown jewel of the business that will continue to contribute robust cash flows. Consensus estimates call for flat EBIT in the HDD head business over the next five years.
I believe that TDK will suffer a significant market share decline in its HDD head business, which when combined with high fixed costs will drive a large decline in earnings. TDK’s current market share is inflated by two long term agreements signed by key customers Seagate (“STX”) and Western Digital (“WDC”) in the wake of industry consolidation and the 2011 Thai floods. Specifically:
Moreover, these minimum purchase agreements were signed when industry volumes reflected a quarterly TAM of 160 – 170mm units, relative to current volumes of 130 – 140mm. In order to compensate for the reduction in throughput and the minimum volume agreements, both STX and WDC are running their internal production facilities at suboptimal capacity utilization levels, resulting in a material drag on margins. This write-up will argue that:
Last quarter, TDK produced an annualized level of 562 million HDD heads. I project this amount to drop ~27% to 410 million based on the factors mentioned above. Due to high fixed costs, I estimate this will reduce segment EBIT from $519 million to $297 million, a reduction of 41%.
Beyond the near term reduction in HDD head earnings, TDK faces a number of additional risks:
I consider TDK an attractive short opportunity.
TDK’s business is primarily composed of three segments: Passive Components, Magnetic Applications, and Applied Film. The Passive Components segment makes capacitors, inductors, and other devices used in a variety of applications from smartphones to cars. The majority of the Magnetic Applications segment consists of the production of HDD heads, with the remainder used in industrial and automotive settings. The Applied Film Products segment produces lithium batteries for consumer electronics and CDs / DVDs. Revenue and adjusted profitability for each segment is shown below:
HDD Industry Overview
Three companies produce HDDs: STX, WDC and Toshiba. Consolidation in the space was completed when Seagate purchased the Samsung HDD business and Western Digital bought Hitachi GST, both of which were announced in 2011. The following table outlines the market as of Q3 2012, with a breakout for legacy production entities (Samsung, Hitachi).
Heads and media represent the key value added components of HDDs. The head, which is essentially a semiconductor, magnetically encodes information on the media, a disc coated with magnetically sensitive material. A suspension assembly holds the head in place, with the entire head and arm assembly called a HGA (Head-Gimbal Assembly). Heads and media govern the information storage capacity of an HDD and allow for modest product differentiation. The two most successful HDD manufacturers, STX and WDC, internally produce the large majority of their heads and media in order to reduce costs, integrate the supply chain, and maintain control of their R&D / technology roadmap.
Manufacturing HDD heads relies on a semiconductor production process, wherein silicon wafer is produced, cut, coated and refined in large fabrication facilities. The highly fixed cost process necessitates large facilities, continuous throughput and high capacity utilization.
TDK is the only merchant HDD head manufacturer in the world. STX, WDC, and Hitachi GST all have internal head capacity and use TDK for a variety of reasons. STX and Hitachi mainly use TDK for swing capacity. WDC uses TDK as a form of outsourced R&D. They purchase cutting edge heads from TDK and as the technology matures, they bring them in-house. Toshiba and Samsung do not have internal head capacity and purchase all their heads from TDK. As mentioned above, recently, STX purchased Samsung and WDC purchased Hitachi. The following table outlines annualized head purchases, as well as TDK’s share at each OEM.
HDD Head Share Loss Thesis
In the last year and a half, both STX and WDC signed multi-quarter, minimum volume purchase agreements with TDK for two entirely different reasons. Unfortunately, the market for HDDs has declined substantially, leaving them obligated to procure a disproportionately large amount of heads from TDK. This has shielded TDK from the weakness in the HDD market temporarily. Conversely, both Seagate and Western Digital are operating their head fabs at low capacity utilization, creating a large drag on their respective cost structures.
Seagate signed a 3 year minimum volume purchase agreement with TDK to satisfy EU regulators in conjunction with the Samsung acquisition. See below for the relevant section from the European Commission Decision dated 10/19/11.
This agreement can be fulfilled by purchasing heads for either Samsung or Seagate brand hard drives. The agreement was signed at a time when industry units shipped (“TAM”) was around 160 – 170mm / quarter, as compared to a current TAM of 130 – 140mm / quarter. At the time, the minimum purchase agreement did not cause concern considering management thought TAM would continue to grow. STX now operates their head fabs at suboptimal capacity utilization to compensate for the purchasing commitment and the smaller TAM. While STX acknowledges all agreements are negotiable, they will likely tread carefully because they are sharing IP with TDK on the development of next generation HAMR heads (the next HDD head technology transition). However, when the agreement expires in 18 months, both TDK and STX will have completed HAMR development and STX will have the opportunity to bring more head production in house.
Moreover, this timeline corresponds with when STX will likely phase out the final Samsung product lines and replace them with STX brand drives. Samsung housed the bulk of its HDD production at its facility in Dongguan, China; STX obtained this facility as part of the acquisition. At this time, the only product of significance produced here is the Samsung M8 mobile drive. Furthermore, SAE Magnetics (a TDK subsidiary) operates the entire facility on a contract manufacturing basis. MOFCOM imposed a variety of restrictions on Seagate as a result of the merger, one of which was the maintenance of Samsung production. The one year timeline recently elapsed and now STX will petition MOFCOM to lift the integration restrictions.
Should MOFCOM approve lifting the restrictions, STX will eventually idle the Dongguan facility – likely in conjunction with the phase out of the M8 product and the expiration of the minimum volume purchase agreement – for a variety of reasons:
- Samsung introduced the M8 in 2011 and HDDs typically carry 18 – 36 month product cycles. STX will likely integrate 2.5” drive technology under a single platform. It does not make sense to continue to allocate R&D dollars to two separate platforms
- Dongguan facility currently underutilized
- Contract manufacturing at odds with STX production strategy. STX likely more efficient and lower cost given core competency in HDD production
- Phasing out M8 will allow for increased utilization of existing STX capacity (especially heads and media).
At two recent conferences, the STX CFO hinted at his intentions for the Samsung head business:
The Thai floods of 2011 severely damaged WDC’s head production facilities. In response, WDC signed a multi-quarter, minimum volume purchase agreement with TDK. For a redacted version of the agreement, see http://sec.gov/Archives/edgar/data/106040/000119312512026854/d275388dex102.htm. As industry TAM declined and WDC rebuilt its head production capacity faster than anticipated, they too currently purchase a disproportionate amount of heads from TDK. The following chart plots the percentage of total WDC heads (ex. Hitachi) supplied by TDK:
WDC hinted at its intention to return to 10-20% external heads on its July 2012 earnings call:
To satisfy EU regulators in conjunction with the Hitachi acquisition, WDC divested certain 3.5” production assets from its Shenzen, China facility to Toshiba. This could potentially offset some of TDK’s share losses at STX and WDC to the extent that Toshiba migrates the heads for this business from Hitachi to TDK. We have excellent detail on the asset sale from the FTC competitive assessment:
- 16 primary production lines at Shenzen, with the optional purchase of 6 more (total 22)
- Estimated annual production capacity of 1.09mm drives per line per year (my own estimate), for total capacity of 18.5 – 24.0mm drives per year
- Lines producing Hitachi Mars product line for non-portable desktops and tower personal computers
- Transfer period of one year (expires May 2013) during which time Hitachi contract manufactures for Toshiba
- Hitachi to continue to supply heads for the business under a 3 year supply agreement, under which Hitachi must be willing to provide heads to satisfy up to 50% of purchased capacity
Based on recent volumes for Hitachi Mars product line, I believe current run-rate volume is approximately 8-10mm drives per year. This is down from ~11mm drives per year at the time of the acquisition. You will note that this almost exactly corresponds with 50% of purchased production capacity and the head supply agreement provisions. I believe this is because Toshiba insisted on having access to heads spec’d in with customers for the life of the current Mars product, and thus signals expected product volumes. In the financial analysis below, I assume that Toshiba replaces these purchased Hitachi heads with supply from TDK, although this likely will not happen for another 2 years.
One should note that my estimates assume Hitachi continues to utilize a lower percentage of outsourced heads than WDC. I believe this makes sense given:
- Desire to fill internal head capacity
- Part of acquisition rationale for HGST was access to strong internal head R&D program
- Higher proportional mix of enterprise drives (which utilize fewer outsourced heads)
Pro Forma Market Share
I assume that STX transitions the Samsung volumes from 100% TDK heads to a mix of 90% internally sourced / 10% TDK. This percentage corresponds with STX’s percentage of externally sourced heads for non-enterprise drives. Since STX uses virtually no externally sourced heads for enterprise drives (more on this later), this figure does not match the 6.1% in the STX row. I assume that WDC reverts back to 15% externally sourced heads, or the middle of their 10-20% target. My conversations with WDC confirmed an intention to lower external purchases below the 17.5% pre-flood average due to the decreased TAM. Finally, I assume 32mm heads per year (10mm drives per year x 3.2 average heads per drive) migrate from Hitachi heads to TDK heads as part of the divestiture of the Hitachi Shenzen 3.5” assets.
The financial attractiveness of migrating to internal head production is compelling given the lower fab utilization from decreased TAM. One expert I spoke with noted that internally sourced heads at STX cost ~50% less than externally sourced heads. If TDK charges an average of $4.17 / head, this implies savings of $2.08 / head and ~$6.00 / HDD, assuming an average of approximately 3 heads per drive. Consider the attractiveness of $6.00 of savings relative to an ASP of ~$50 for a consumer HDD.
The following table estimates the impact on TDK’s revenue and EBIT from the projected market share loss. I assume a variable contribution margin of ~35% for the lost heads. One expert I spoke with noted the back end of the HGA production process (wafer production) has variable contribution margins approaching 70%. I have counterbalanced this with the production of the arm and the final assembly, which has a more variable cost structure.
Toshiba Exposure Risk
TDK currently sells 39% of its heads to Toshiba (I expect this to grow to 62% based on market share losses elsewhere), an industry player that is structurally disadvantaged relative to STX and WDC. Toshiba has the smallest market share in the industry (~15% as compared to over 40% for both STX and WDC), giving them less business to leverage their fixed costs and R&D over. Likewise, they have no internal heads or media production, leaving Toshiba the structurally highest cost manufacturer of HDDs. Finally, Toshiba holds by far the smallest market share in enterprise HDDs, leaving them exposed to the PC and notebook markets. With a structurally disadvantaged cost structure and little enterprise exposure, I expect Toshiba to continue to find it challenging to compete against STX and WDC. This does not bode well for TDK, considering TDK supplies 100% of Toshiba heads.
PC and Notebook Risk
HDD “bulls” often point to the argument that while SSDs may cannibalize HDDs (whether from tablet growth or SSD replacement of HDDs) in the consumer market, the total amount of data to be stored continues to grow rapidly. In other words, bulls argue that regardless of the medium of data consumption, data growth continues unabated, and data must be stored somewhere. Even if local storage on PCs and notebooks stagnates, many sources estimate aggregate data growth rates of ~40-50% per annum, implying significant storage needs in the cloud and data farms. For mass storage uses, SSD is not nearly cost competitive to HDDs on a price/GB basis.
Unfortunately, TDK has little exposure to this growth opportunity. In fact, TDK’s core expertise lies in the production of heads for laptops, one of the markets most likely to be cannibalized by tablets and SSDs. As shown below, approximately 75% of TDK’s heads are used in mobile applications, as compared to roughly 50% for the industry as a whole.