Data Domain (DDUP) is a Greylock and NEA-funded startup that IPO’ed in June 2007 and has since grown rapidly selling a bundle of hardware and software that helps companies do data backup.  At over 200x consensus ’08 earnings and 5x consensus ’08 revenues, the valuation has massive expectations for growth and bottom line improvement built into it.  The street view is that competition trials far behind Data Domain.  However, in fact the competitive situation has changed drastically in the past 6 months.  Whereas DDUP was first to market with “data deduplication,” now nearly all the incumbent leaders in backup storage have systems that include deduplication.  The emerging consensus in the industry is that data deduplication is not a standalone product but simply a feature to be bundled with backup disks or backup software.  As competitors start to gain traction in the next 6-9 months, DDUP should lose market share, suffer margin compression, and it is likely the street will reevaluate its growth potential.    At 30x consensus ’09 earnings, the stock would trade at under $11.  Management is not sticking by the stock—they have sold 950,000 shares in 2008, with nearly 400,000 shares sold by Frank Slootman, the CEO, since May!

Deduplication (“dedupe”) solves a key problem in data backup: by eliminating duplicate data during the backup process it frees up 10-20x (or more) additional space on backup disks, making them price competitive compared to tape.  For this reason the market is moving away from tape backup, which is slower and less reliable than disk backup.  Dedupe can be done either on the backup server (“source-side”) or on the server that manages the backup disks (“target side”).  The best-selling solutions in the market backup data “on the fly” as the data travels to the backup target (known as “in-line” systems), compared to systems that deduplicate data already stored on disks (“post-process” systems).   Finally, the dedupe can be done at the file-level, at the block level using fixed size blocks, or at the block level using variable-length blocks: variable-block dedupe tends to reduce the data size the most.

Dedupe has been used in the storage industry for years.  According to one industry participant, it is “a tough space to protect,” since the hash functions that are used for dedupe are public domain.  Data Domain’s innovation was around a certain implementation of the hashes to do inline data deduplication, and this is what they have patented (see USPTO #7,373,464, and 7,305,532).  Quantum, a leader in tape-based systems, has a patent covering variable block length deduplication, which is what forced DDUP to sign an IP-sharing agreement with them in March ‘07.  Data Domain’s intellectual property has not stopped all the major storage vendors from bringing dedup-enabled products to market in the last few months.

Gartner estimates that Data Domain currently has 60% of the market for target-side dedupe-enabled backup systems, with consensus analyst estimates of $256 million in revenues for them for 2008.  They are followed by Diligent, Quantum, and ExaGrid.  However, the competitive situation has changed drastically in the past 6months:

IBM, EMC, and HP are the three global leaders in storage disks, with a combined market share of over 50%.  IBM and EMC are also leaders in backup software.  They have established relationships with customers which they can leverage, global service organizations to reassure customers, and as they come to market, they can bundle dedupe with their market-leading storage disks.  Compare this to Data Domain which sells 86% of their product through indirect channels, has an outsourced service organization, and bundles its software with commodity disks.  EMC’s new product series, powered by Quantum’s deduplication technology, matches Data Domain’s line of products nearly feature for feature.  Due to Quantum’s patent cross-licensing agreement with Data Domain, the technology offers both post-process deduplication and the inline deduplication that has proven so popular on Data Domain’s products.  It’s notable that Quantum’s new DXI 7500 series, which features the same technology, has been getting good reviews in the industry.  EMC’s product runs Quantum’s dedupe technology on EMC’s popular Clarion disks.  In sum, Data Domain’s competitive moat is thin.  Gartner estimates that Data Domain will have 25-40% share at the end of next year—a result that would be disastrous for Data Domain’s valuation.

Indeed, the emerging consensus in the industry is that data deduplication is not a standalone product but simply a feature to be bundled with backup disks or backup software.  NetApp, the leader in NAS storage, has taken the approach of bundling a lower-quality “good enough” solution with its drives for free.  Released in April ’07, NetApp’s dedupe solution has grown rapidly to an installed base of 3,700 licenses in March ’08 (Data Domain reported 1,800 customer in the same month).  Finally, the leaders in data backup software are also putting pressure on Data Domain.  Symantec partnered with NetApp to do joint development on a product named SnapVault, so it integrates seamlessly with NetApp’s dedupe solution.  CommVault offers single-instance-storage today and is planning to add a block-level dedup feature in Q1 ’09.

Accelerating competition will take a toll on DDUP’s margins.  According to one industry expert, pricing on dedupe systems started two years ago at 0-10% below list price, and then fell to 10-15% below list as smaller competitors came on the market.  Looking forward, he now advises clients to push for 30% below list price.  Some people even talk about the potential for some products to fall to 40-75% below list price, as vendors like EMC and Quantum cut prices to take market share from Data Domain.  I assume that pricing will fall by 10% in ’09 and again in ’10, putting severe pressure on Data Domain’s profitability as their gross margins fall from 72% in Q2 to 70% for ’08, 60% for ’09, and ‘50% in 2010.

Data Domain has set its long-term target to grow its deduplication product line to a $1 billion business.  However, that figure is for a bundle of deduplication software as well as related hardware, including servers and disks.  So, it’s misleading to talk about a market for “deduplication”; rather, the market is for hardware and software sold to companies that need to do data backup.  Gartner estimates that this will be a $1 billion market at the end of next year.  A former product manager at EMC estimates the long-term potential of the Data Backup market to be $2-3 billion.  These are large figures, but far too small to accommodate the growth expectations embedded in DDUP’s stratospheric valuation. What’s more, although the trend in the market is clearly for huge growth (since enterprise data needs are increasing quickly and companies are transitioning from disk to tape backup) the majority of the revenues will come from hardware, especially disks.  IBM, HP, EMC, Dell, NetApp, Hitachi, and Sun already dominate the global market for storage disks.   Another factor to consider is that economic weakness could reduce enterprise IT budgets next year across the board.  In this environment, companies may choose to delay their transition from tape to disk.  In DDUP’s Q2 ’08 call, management notes particular strength in the financial service vertical which is one of the industries where IT budgets are likely to suffer the most in the following quarters.

Adjacent to the Data Backup market are the markets for Data Archiving and Disaster Recovery.  They are by definition smaller markets than Data Backup: nearly every enterprise backs up data, but fewer do archiving and disaster recovery.  Archiving is a specialized market, primarily targeted to helping companies meet regulatory requirements.  EMC has had a robust presence in this market for years, and the product needs are different from the Backup market.  For instance, file-level deduplication is preferred to block-level, since companies like to delete specific files at various times, after the regulatory requirements to hold those files have expired.   Disaster Recovery is a better fit for Data Domain, and they already have products to meet this need.  However, even in this market, companies like EMC have offered things like “synchronous copy” for a long time.  Overall, it is unlikely that adjacent markets will be large enough to meaningfully effect Data Domain’s potential available market.

RISKS

The major risk to the short thesis is that Data Domain could strike a licensing deal with a major storage vendor.  Dell, Sun, and Hitachi would be the most likely options, since all the other major vendors already have deals in place.  However, each of these vendors have under 10% market share in disk storage.   So, it’s hard to see how a licensing deal could have a significant enough effect to justify DDUP’s valuation.