Comment Violin Memory, the flash array start-up, has snagged $40m in C-round funding, just three months after a $35m B-round, signalling an acceleration of its development activities: it's going to add more flash strings to its bow.
Violin supplies externally attached flash memory arrays, with a 10TB, single level cell, 3200 performance product, and the 40TB 3140, a capacity-optimised, 2-bit multi-level cell product.
The new money comes from "several of the largest public companies in the computer industry" (both announced and unannounced). Toshiba and Juniper are two announced investors. There has been a lot of recent co-operation with HP helping that company compete with Oracle's Exadata appliances.
Don Basile, Violin Memory's CEO, said: "Our new capital will allow us to speed up product development and global expansion in Europe and Asia to surpass $100m in revenue this year, the first step in building a billion dollar company. This pace of revenue growth will continue to make Violin the fastest growing storage company in the last decade."
That's a high ambition for a start-up company and implies that Violin feels it can do this despite the first mover and market-leading advantages of Fusion-io - which Don Basile used to run.
We asked Basile some questions about the new funding. He reckons a "silicon storm" – a flash flood perhaps – is going to change data centres for ever. He says that recent events, such as Larry Ellison talking about running Oracle software on SuperCluster using 138 flash memory; arrays; EMC's flash strategy announced at EMC World; and the Fusion-io IPO have raised everyone's awareness of flash memory's potential.
Basile said: "In the creation and announcement of our B-round we had already created the C-round and had strong interest. The focus of the last few weeks on EMC, Micron, Intel and what appears a successful listing of ... Fusion-io, has focused attention on the space in general."
The time is ripe to strike now and Violin, focusing on the primary storage market, is boosting its product and channel development efforts. Basile said there is more money available for this than we realise: "In addition to the $40m of equity, we have arranged over $140m of debt and credit lines to support our rapid manufacturing expansion. So while not going public ourselves yet, we have access to public level financial support to build the business."
Basile says business looks great: "The Memory Array we unveiled last May and shipped in the last six months of 2010 ... garnered double-digit revenue for 2010, and will garner triple-digit revenue in 2011."
He claims: "This revenue ramp is faster than all the recent wins in the space – 3PAR, Data Domain, EqualLogic, Isilon, and Compellent."
There is no reason this can't continue as the total addressable market for Violin products is huge. This is the way he describes it: "The Micron, STEC, and Intel cards should make the [PCIe] card space highly contested. In viewing the Fusion-io [IPO] S1 document, if you subtract the Facebook (nearly 50 per cent) and large customer number two (20 per cent) revenues, you are left with under $36m of revenue. At $3,000 per card this is just 12,000 cards over 2+ years.
"The three server OEMs – HP, IBM and Dell – ship together on the order of 9 million servers a year. So the attach rate to date ... has been a rounding error. The server-side flash market remains wide open."
Limited Samsung and Toshiba advantage
Basile thinks Micron, Samsung and Toshiba have many advantages, especially when the product is a component card or drive, but: "These vendors ... have shown little expertise in systems and software, the domain of Violin Memory. Toshiba and Samsung dwarf the rest of the industry [meaning] Intel, Micron and Hynix, in volume and investment in NAND. These two have the ability technically and financially to drive the adoption curve faster and tune the NAND through the 2x and 1x [geometry] transitions for enterprise use."
He says: "All this requires more engineering and more investment to stay relevant, and will make it increasingly harder for new entrants and small entrants to compete. Intel [has] the greatest potential in the space. It has been surprising [that] it has taken so long for them to produce a viable card platform. However their NVMHCI initiative on card drivers, and their long-term advantage around the x86 chip set, make them a favourite on server-side flash solutions long term. Their small fab capacity however, may limit their ability to play in the broader market as they defend against ARM."
Basile says he believes he can build Violin Memory into a billion-dollar-a-year revenue business. He has the financial resources to do this, his technology IP has increased with the Gear6 acquisition, and he appears to have a great desire to overtake the company that ousted him from its CEO position – current flash product market darling Fusion-io. It all adds up to exciting times in the flash industry as suppliers believe there is everything to play for. ®