XtremIO makes storage arrays based on flash memory chips, and is a would-be rival to Violin Memory, the Silicon Valley start-up that’s revving its engine for an IPO later this year, following an $80 million Series D funding round which AllThingsD reported exclusively last month.

Another player in the all-flash storage array business is Pure Storage, which came out of stealth mode last August with a $30 million Series C led by Redpoint Ventures.

News of the deal gave shares of Fusion-IO a jolt. Fusion-IO rose 50 cents, more than 2 percent, to $21.63, just as the markets opened for trading in New York. As of yesterday’s close, Fusion shares have fallen by more than 6 percent since its IPO debut last June.

Fusion is a founding member of the Flash Madness Club. Its flash memory insert cards for servers are widely used in data centers of companies like Apple, Salesforce.com and Facebook, speeding up the ability of servers to process data by eliminating bottlenecks created by conventional hard drives. Its customers also include Hewlett-Packard, Dell and IBM among other server manufacturers.

That funding round, I’ve since learned, was so oversubscribed that it reached $80 million and now includes a significant new investor: GE Asset Management. A filing is expected with the U.S. Securities and Exchange Commission Tuesday.

GE Asset Management is joining a funding round that includes strategic stakes from Toshiba, the Japanese chip and electronics maker, and networking gear player Juniper Networks as well as Highland Capital and SAP Ventures, the investment arm of German software giant SAP.

The funding is Violin’s latest step toward filing for an initial public offering. Violin CEO Don Basile told me that the company has selected four banks to work with on the offering, following the bake-off process we mentioned last month: J.P. Morgan, Deutsche Bank, Bank of America Merrill-Lynch and Barclay’s, confirming a report from Bloomberg News last month.

Basile told me that he expects Violin’s road show will take place during the summer and that the company is now well within what he says is a 180-day window during which it will go public. That would place the offering no later than October 27.

It would make Violin the second company using flash memory in the data center to go public within roughly a year. Fusion-IO went public last June in a successful offering that boosted the company’s valuation above $2 billion.

And if the interest of pre-IPO investors is any indication, and it often is, Violin’s public offering, whenever it finally does occur, should prove popular.

One early customer is Rob Legato, the visual effects supervisor who won an Academy Award for his work on the Martin Scorsese-directed hit motion picture “Hugo.” Legato will be talking about ioFX with Fusion-io chief scientist and Apple co-founder Steve Wozniak at a conference in Las Vegas next week.

Fusion does some cool stuff with flash memory. Here’s the part where I roll out the old metaphor that has served me so well: In pretty much any computer, you can think of the processor as a fast-moving, highly efficient, type-A personality, constantly in a hurry, and always waiting impatiently for the rest of the system to give it more work to do. The slowpoke in the deal is the hard drive, which, though it’s already spinning at a super fast rate, just can’t get data to the processor fast enough. So the processor sits around, tapping its foot and looking at its watch, waiting for the other parts of the system that feed it data to work to keep up.

In high-performance computing, where there’s more data to be crunched than in most average computing situations, this is sort of a big deal. You want the processor to be as busy as possible — mainly because the systems are so expensive, and you want to get your money’s worth out of them — but also because jobs get done faster.

So Fusion-io’s stock in trade is a series of insert cards that bring flash memory right up next to the processor. The flash chips grab great big armloads of data and hold on to it, handing it off to the processor in a way that keeps it happy and busy and not impatiently waiting — at least not so much.

We’ve seen the technology brought to bear at places like Credit Suisse, which added Fusion’s flash cards to its trading systems. And its technology is also used in data centers belonging to Facebook and Apple.

On top of that, Fusion has relationships with all the big server vendors: Hewlett-Packard, IBM, Dell and SuperMicro all sell systems with Fusion-io on board.

Workstations are essentially heavily tricked-out PCs that are used primarily in two professions: Animation and special-effects work for movies and TV and computer-assisted design and modeling, used by folks who design buildings and cars and planes and pretty much anything else you can think of. They have the same problem that servers have — agitated processors constantly waiting for the rest of the system to catch up with them.

At this point, none of the workstation vendors are offering the card as an option, but if you’ve got a professional workstation — like, say, an Apple Mac Pro, which has three PCI Express slots — you might add one of these cards and speed up your work. In the meantime, the company is working with workstation vendors to get the ioFX insert cards certified. My guess is there will be more than a few visual artists who won’t bother to wait.

Fusion-io shares are up almost 11 percent — or $2.64 — to $27.30, as of 11 am ET; not so much on this news — workstations are kind of a low-volume market — but on an analyst report from Piper Jaffray suggesting that Cisco Systems may be close to a deal to add Fusion-io’s flash technology to its Unified Computing System platform.

The report goes on to suggest that Cisco could, over the next three or four quarters, become one of Fusion’s bigger customers, along with Facebook and Apple, and could account for more than 10 percent of Fusion’s business — which could, in turn, lead to a doubling of revenue this year. For the record, sales were $197.2 million in Fusion’s fiscal 2011. Do the math.

The funding round includes strategic stakes from Toshiba, the Japanese chip and electronics maker; networking concern Juniper Networks; and funding from new investors, including Highland Capital and SAP Ventures, the investment arm of German software giant SAP.

Violin CEO Don Basile also told me today that the company is in the process of picking bankers that will likely lead it to an initial public offering before the end of 2012. “We had our final bake-off last week,” he told me, though he didn’t disclose who had won it.

That Violin was raising capital was disclosed in a filing with the US Securities and Exchange Commission. A formal announcement on the funding will come Monday.

Violin has been growing pretty aggressively in recent months. Basile told me that the company now has 320 employees, up from 50 in the last six months. It has been building up a global sales force with 40 people working in Europe and the Middle east. That team is run by Garry Veale, the former head of HP’s Storageworks operation in Europe. Earlier this month it hired Martin Darling, a former EMC sales exec to run its sales team in Asia.

Basile says the investment will be used press down on the gas pedal and keep growing, but also to look seriously at an IPO before the end of 2012. “The funding gives us the means to grow as a private company, but also to look at the public markets if the conditions are right,” he said. “It’s more likely than not that we’ll be a public company by the end of the calendar year.”

Narayan Venkat has joined Violin as VP of product management. He spent just a bit more than a year at VMware, where he led its storage initiatives. His resume includes time at chip companies including LSI and Intel.

At Violin, he’ll be in charge of pushing Violin’s flash technology into the data center. As I told you last summer, when Violin raised $40 million from Toshiba and Juniper Networks and several individuals, its flash arrays run faster than old-school storage arrays, while reducing both the physical footprint needed for the hardware and the power consumption. Hewlett-Packard resells its gear, and AOL is a big customer.

“Flash is the biggest disruption in the data center to come along in years,” Venkat told me. Violin’s last big hire was its CTO, Jonathan Goldick, who also came from LSI.

Shares are down more than $4, or about 13 percent. The freakout appears to be coming from gross margins that shrank to 51 percent from almost 59 percent in the prior quarter, and despite the fact that sales more than doubled sequentially to $84 million from $31 million before.

CEO David Flynn called me up a little while ago to talk about the results, and he reminded me that Fusion launched its new IO Drive 2. It’s a transition to a new product line that’s proving tricky. New products built on new technologies are always a little more costly to build up front, and that’s compounded by the fact that early adopters, when they buy the new stuff, take the lower-end version and not the more expensive and more profitable one.

Also, enterprise customers who buy the new stuff are always conservative and take longer to decide whether they want to buy it or not, he says. Even so, the company has sold 10,000 of the new drives.

But? There’s a new customer of record: Salesforce.com is now a Fusion-io customer, and has joined the likes of Apple and Facebook, which is using the flash-based chips in the servers running in its data centers around the world.

And Salesforce isn’t buying it directly from Fusion, but rather through one its OEM partners, which include Hewlett-Packard, IBM and Dell, though Flynn wouldn’t tell me which one it is.

Salesforce is one of six customers who bought more than a million dollars worth of Fusion’s stuff this quarter and of those, four were repeat customers, Flynn told me.

The Salesforce win is also important, Flynn says, because some have wondered whether Fusion’s technology, while popular with high-end enterprises like banks and Facebook, would make sense for applications that tend to be used in mid-tier businesses, which Salesforce’s mainline CRM application often is. The lower end of the enterprise software market is moving toward cloud-based software, which is often referred to as Software as a Service, or SAAS. “By helping those companies, we are indirectly driving business in the mid-range of the market. Apple and Facebook are in the SAAS business too, it’s just that their customers are consumers.”

One interesting fact that Flynn shared with me: His first job out of college was working for Oracle. His boss at the time? One-time Oracle exec and now Salesforce CEO Marc Benioff. A small world it is, indeed.

In what can only be described as a strange collision of two distinct yet oddly similar universes of the geek canon, Steve Wozniak — Apple co-founder, friend of Steve Jobs, and chief scientist of chip memory concern Fusion-io — will have a conversation at Thursday’s DEMO conference with Leonard Nimoy, the actor famous for playing Spock in the “Star Trek” TV and film series.

Fusion just announced the pairing on its corporate blog. The company says the two will “share their thoughts on technology’s past, present and future.”

It’s probably no surprise that Wozniak was a fan of the original 1960s vintage TV show in his early adulthood. In a speech he delivered at a conference earlier this year he said that during his days working at Hewlett-Packard designing calculators, he’d come home from work “watch ‘Star Trek,’ eat a TV dinner, and do electronics projects.” So Woz will probably be thrilled to hang out with Nimoy, who’s always been a favorite among the “Star Trek” fan community.

It’s a little hard to guess precisely what all this will have to do with Fusion-io’s flash memory technology, which essentially speeds up conventional servers by adding an extra layer of memory to keep data close by the processor so it doesn’t stand around waiting for the hard drive to catch up. Nimoy is, however, an old hand at talking about how consumer technology that was science fiction on the TV show — mobile phones are essentially “Star Trek” communicators, for example — is now a reality. (See the video below for an example of that.) I’m sure it will be fun to see regardless.

One thing that will happen: A geek swarm on Foursquare. Woz is active on Foursquare, so expect lots of his followers to check in all at once and maybe trigger a swarm.

Apple isn’t commenting and is officially treating all this as rumor and speculation (it rarely comments on acquisitions, anyway). But the deal is being reported in Israeli newspapers, and the office of Israeli Prime Minister Benjamin Netanyahu tweeted a welcome message to Apple today, which sure feels like confirmation. So I’ll proceed under the assumption that the reports of this acquisition are true.

Welcome to Israel, Apple Inc. on your 1st acquisition here. I’m certain that you’ll benefit from the fruit of the Israeli knowledge.

December 20, 2011 3:55 am via webReplyRetweetFavorite

@IsraeliPM

The PM of Israel

That makes this a cause for celebration. With the Anobit buy, Apple is now the latest member of the Flash Madness Club, which I created over the summer, in the wake of the Fusion-io IPO and other activities by notable flash-technology companies like Violin Memory, Pure Storage and Qwilt.

So why is Apple willing to throw down a reported half-billion dollars on this company? It’s because flash memory has a fundamental problem: As it ages, its ability to store data wears off. This problem is sometimes compared to the semiconductor equivalent of Alzheimer’s. Individual cells on the flash-memory chip lose their ability to store the individual ones and zeros that make up the pictures and music and other data they may be storing, especially after millions of read-and-write operations — the act of putting data on the chip and then loading it from the chip for use. After a lot of heavy use — this can vary depending on the chip — the chips begin to suffer problems with “endurance.”

As flash starts to show up in data centers and PCs and other places beyond consumer gear like iPhones and iPads, this becomes a more important problem. If your iPad gets old enough to suffer data-endurance problems, it’s a pretty simple matter to replace it. But in the more rigorous world of an enterprise data center, where millions of reads and writes will be done on a chip daily, data endurance is a potentially very expensive problem. In the enterprise, a solid-state drive is considered suitable only if it can stand up to five full-drive write cycles, where the drive is filled to capacity and then erased every day for five years.

Anobit’s solution to these problems involves techniques known as memory-signal processing and the use of some secret-sauce memory-processing error-correction algorithms, plus some management tricks for moving data around a flash chip in more efficient ways, in order to make them last longer.

It’s also the sort of problem that a company like Apple — which is the world’s largest consumer of flash memory, and has been for several years — would want to solve. Think of the many places where Apple uses flash — the iPad, iPhone, iPod, MacBook Air and Apple TV. And those are just the products we know about, so far. Flash can’t help but appear in many more products.

On top of that, flash technology plays a significant role in Apple’s data centers. Fusion-io, the company that builds flash-based insert cards that speed up garden-variety servers, has named Apple as a significant customer, so there’s plenty of flash inside Apple’s facilities in North Carolina. Flash endurance can’t help but be a problem Apple might face with its iCloud service, for example.

Israel has a big connection to the flash industry. SanDisk’s founder, Eli Harari, is Israeli; a few years back SanDisk acquired an Israeli company called Msystems, which, if my memory serves, was the first to popularize what we now call a thumb or keychain drive. So, historically, there have been a lot of useful innovations on flash memory that have come out of that country. Supposedly, the deal calls for Apple to open a research center there, so it will get the benefit of ongoing innovations on flash. Chances are it’s going to need a few.

I’ve asked Apple for input; last week, when news of the potential deal first surfaced, the company declined to comment.

If this is a done deal, it would follow an M&A pattern that Apple has pursued for some time. Though the company has enormous cash reserves, it has generally made smallish (by its standards) deals for technology vendors it is already using — in this case, it is using Anobit’s flash memory chips in its iPads, iPhones and MacBook Airs.

But Apple spends much more cash — billions a quarter — locking up inventory for the raw components than it uses in its gadgets, like flash memory or display screens.

Fusion-io shares closed at $38.10 — up 9.17 percent — during the regular session, and continued to climb by an additional 1 percent in after-hours trading. The shares have increased by more than 100 percent since they debuted on the New York Stock Exchange at $19 early this summer.

The main news came in the form of a new product, and the publication of news that Fusion-io technology was used in a high-performance computing project at the Lawrence Livermore National Lab.

People tend to think of Fusion-io as building traditional storage, but its main mission is to get data closer to the processor in a server, so that that processor doesn’t have to sit around waiting. Processors are super speedy and super impatient. Think of the processor as the impatient Miranda Priestly — played by Meryl Streep in “The Devil Wears Prada” — and how Anne Hathaway’s character, Andy Sachs, is never fast enough for Priestly about handing her something she needs right away. Microprocessors hate nothing more than waiting for a hard drive to serve up the data they need.

Fusion-io’s drives try to speed that process up — and make microprocessors happier — by using flash memory built into an insert card and installing it close to the processor in a system. The news, announced at the Supercomputing conference in Seattle today, is that Fusion-io debuted a 10 terabyte version of its high-end ioDrive Octal product. You can now pack four of these into a single server, and have 40 terabytes of data right up close to those impatient processors. Companies like Hewlett-Packard, IBM, Dell and Supermicro build Fusion-io’s products into their own products.

The other news also had a supercomputing wrinkle to it. A machine that Lawrence Livermore called “Leviathan,” packed with Fusion-io cards and Intel processors, broke a record in processing a graph with more than 68 billion nodes. Well, it didn’t just break the record, it shattered it, as that number of nodes in a graph is four times the prior record. What that means, in English, is that the computer plotted a mathematical graph with more than 68 billion points of data.

Apparently — and I’m just learning this now — there’s a separate version of the Top 500 list called the Graph 500 which focuses on simulating 3-D problems.

It’s a lot to take in, but the main point is that Fusion-io seems to be showing that it has a going business. Critics of the company have argued that it relies too heavily upon its biggest data-center customers like Facebook and Apple, and that it will be vulnerable to slowing sales when those companies are through building their infrastructure. The problem with that argument is that there’s always another impatient processor throwing an impatient diva fit while waiting for data.

Also, I should note that today’s 9 percent move comes after Fusion shares fell about the same amount on word last week that the company is planning a $350 million secondary offering. When investors heard about that last week, they sent the shares plunging by more than 8 percent, territory it has since reclaimed. It has been a bumpy, volatile ride for Fusion-io, no doubt. In the five months since the debut, the stock has traded as low as $15, and almost as high as $40. That’s IPO investing for you.

I was on the phone with its two founders: Alon Maor, CEO; and Dan Sahar, VP of marketing. They had just started telling me about how they plan to sell network appliances that network operators — like, say, Comcast or Time Warner or Verizon — might put on their network in order to help them meet the growing demand for video content. The aim, Maor told me, is to get the most popular content as close as you can to the customer.

The first thing that popped into my mind was creating an appliance that sits on the network; close to, but not in the customer’s house. Maybe in the nearest network hub or central office. It turns out I was right. Then I wondered aloud what Qwilt might be using as storage technology. Could it be, maybe … flash memory? The chips that have so revolutionized the data centers of companies like Facebook and Apple and the banking systems of Credit Suisse, among others, when put to use by the likes of Fusion-io and Violin Technology?

Why yes, it does use flash memory, they told me, making them the latest member of the steadily growing “Flash Madness” club, which gives me yet another excuse to use the image taken from the cover of Flash Comics #1, circa 1940. For reference, the other members are Fusion, Violin Memory and Pure Storage.

Maor and Sahar laughed on the other end of the line at my guesses. “Would you like a job in our engineering department?” Sahar kidded me. I didn’t answer, because I wasn’t done guessing things like how Qwilt does what it does. “You must use some kind of algorithm to figure out what’s popular,” I said. Right again, mostly. The interview hadn’t been going for as much as five minutes, and I hadn’t even asked a single question and pretty much had it all figured out.

Well, not everything. There was the small matter of funding. Qwilt has raised $24 million in two rounds from Accel Partners, Redpoint Ventures and the Crescent Point Group, a fund based in Singapore. Maor is a Cisco veteran who got absorbed into that company following its $200 million acquisition of P-Cube. Before that, he was an engineer at Seabridge, which is now known as Nokia Siemens Networks. Sahar was director of marketing at Crescendo Networks, now part of F5 Networks. Tom Dyal, a Redpoint partner, is on Qwilt’s board.

Video is so popular with consumers that Internet services providers are struggling to get their networks scaled up to meet the demand, Maor says. The traditional way to solve that problem when everyone is watching the same show on Hulu, or the same movie on Netflix, is to just add routers and pray. That’s expensive. What if you could add some extra piece of gear that works with the existing network infrastructure? If you could figure out what was the most popular show in a particular neighborhood, make a copy of it right in that very neighborhood, and deliver it from there rather than all the way back from Hulu’s or Netflix’s data center, you’d lessen the network’s burden.

So that’s exactly what Qwilt does: It has three patents pending on processes for determining what video applications are being used on a network, and for figuring out what content is most popular in a particular area. So if you’re in a neighborhood full of “Jersey Shore” fans, the Qwilt box would figure that fairly quickly, and keep copies of it close at hand so that everyone gets their required daily dose of Snooki.

Also on Qwilt’s board is Rich Wong of Accel; Peter Wagner, an independent board member who has previously worked at Accel; Ohad Finkelstein, a partner at Crescent Point; and Giora Yaron, the former chairman of Mercury Interactive, which is now part of Hewlett-Packard. Also investing is Rob Glaser, the former CEO of RealNetworks.

Got all that? I told you it was an easy interview.

Inside, the phone is similar too, but there have been some strategic changes from one generation to the next that have important implications for Apple’s many suppliers. According to a teardown analysis conducted by the research firm IHS iSuppli, chipmaker Intel, which last year acquired the wireless operations of the German chip concern Infineon, has been almost entirely bounced out of the 4S in favor of a set of chips from Qualcomm. The shift to Qualcomm had been rumored as far back as last September.

Before Intel acquired its wireless unit, Infineon had previously supplied Apple with a chip known as a baseband processor that Apple had used in combination with chips from Skyworks and Triquint to work with wireless phone networks. “Qualcomm is the big winner here,” says Andrew Rassweiler, an analyst with IHS iSuppli who conducted the teardown. “It is selling Apple a whole suite of chips that adds up to about $14 to $15 per iPhone.”

Intel spent $1.4 billion to acquire Infineon’s wireless chip operations last year in a move seen as meant to shore up its presence in the wireless phone industry overall. It has struggled to win business for its Atom line of microprocessors, which are aimed at mobile devices like smartphones and tablets.

Infineon still has a small chip in the iPhone, but Rassweiler says it’s far less significant and a lot less costly than the one it supplied Apple before. “It’s almost like Apple threw them a bone with a 50-cent part after they lost a much more high profile chip that cost about $10,” he says. Intel had no comment.

ISuppli regularly conducts teardown studies of wireless phones and other consumer electronics devices in order to find out who a manufacturer’s vendors are — like most manufacturers, Apple prevents its suppliers from identifying themselves, much as they’d love to — but also to determine what each part costs. The combined cost of components — analysts check on the list prices of each part — is known as a bill-of-materials (BOM) estimate that gives a fair idea how much a manufacturer, in this case Apple, makes in gross margin on each device sold. Apple doesn’t disclose its gross margin on a per-product basis but when it reported its quarterly results yesterday it said its overall gross margin was 40.3 percent.

In the case of the iPhone 4S, Rassweiler estimates that the BOM cost ranges from $188 for the 16 gigabyte version of the iPhone 4S to $207 for the 32GB version and $245 for the 64GB version. Apple and its carrier partners sell the phones for $199, $299 and $399 respectively, typically with a two-year contract for wireless service that carriers use to subsidize the cost they pay Apple.

The costliest components are the ones that determine the price: Memory chips. Apple has been known in the past to rely mostly upon South Korea’s Samsung Electronics, the world’s largest supplier of memory, and from Japan’s Toshiba. In the phone that Rassweiler’s team tore down, the memory chips came from Samsung rival Hynix Semiconductor. “That struck us as a bit of a surprise,” Rassweiler says. It’s hard not to wonder if adding Hynix to the stable of iPhone memory suppliers is a partial response by Apple to the complicated patent fight it is waging with Samsung in courtrooms around the world.

Even so, Samsung appears to be have maintained its role as the manufacturer of the Apple-designed A5 processor that provides the iPhone 4S, and also the iPad 2, with most of its computing horsepower. Some published reports in recent months had suggested that because of the patent fight, Apple might end a relationship that dates back to the original iPhone and move its chip manufacturing contract to Taiwan Semiconductor Manufacturing Co., the huge chip manufacturing foundry. Rassweiler says there’s no sign on the latest A5 chips that that has occurred. “The markings are the same as what we saw in the iPad 2,” he says. The estimated cost for the A5 chip is $15 each, he says.

Apple started designing its own chips for the iPhone and iPad products beginning in 2010 with the release of the first iPad. The chip is thought to have been designed by teams from PA Semi and Intrinsity, two privately held chip design firms that Apple acquired in 2008 and 2010 respectively.

However, it’s also clear that the A5 chip is taking on more of the heavy computing lifting inside the device than the previous A4 chip, Rassweiler says. For example: The iPhone 4 contains a chip from privately held Audience Semiconductor, based in Mountain View, Calif., that handled noise cancellation. There’s no such chip inside the iPhone 4S, Rassweiler says, so it appears that noise-cancellation duties may have been moved to the beefier A5 chip itself.

Triquint Semiconductor provided a set of chips that make up a wireless transmit module that works with the wireless phone networks. Triquint has traditionally been an iPhone supplier, Rassweiler says, but the value of what it supplies to Apple appears to have dropped. One wireless chip company that has seen the value of what it supplies to Apple increase is Avago Technologies. Like Triquint, it too has been an iPhone supplier, but the overall value of the chips it supplies has gone up in the 4S.

STMicroelectronics, the European chipmaker, maintained its role as the supplier of gyroscope chips that help determine the phone’s position and rotate the screen for playing games and displaying pictures and videos. AKM Semiconductor again supplied the compass chip. Texas Instruments continued in its role supplying the chip that controls the iPhone’s display, and an audio chip.

One vendor could not be identified. Rassweiler says that Apple appears to have taken pains to hide the identity of the company that supplies the parts that power the iPhone 4S’s highly regarded 8 megapixel camera. This is not new, and the candidates include Largan Precision Co., a Taiwanese supplier of camera modules to wireless phone companies, and Omnivision. “We don’t know exactly who makes it,” Rassweiler told me. Whoever the supplier is, Rassweiler estimates the camera added $17.60 to the cost to build the iPhone. And they’re likely to make a lot on the deal. IHS iSuppli is forecasting that Apple will sell 81 million iPhone 4Ss around the world next year.

Update: A few of you have written in saying that it was Sony who supplied the camera. Maybe. The folks at Chipworks dissected the camera module and found a Sony-made CMOS image sensor inside it. That doesn’t make the whole module a Sony’s however. It could be a Sony camera or it could be that whoever made the camera used a Sony sensor. And last week Barron’s reported on some debate among analysts over whether or not Apple has split the camera supply contract 50-50 between Omnivision and Sony.

SanDisk will sell co-branded four gigabyte and 8GB memory cards that are similar in features to the company’s current entry-level products.

Though the partnership is initially centered around SanDisk selling co-branded Eye-Fi chips in Europe, CEO Yuval Koren notes that it opens the door for the two companies to work together more closely down the road.

“We really see this as the beginning of a deeper partnership,” Koren said in an interview, noting the deal has been some time in the making.

Although Eye-Fi had been distributing its memory cards in a few European countries, such as Germany, France and the United Kingdom, its efforts had been limited in scale.

“They are obviously going to take a much broader approach to that,” Koren said. With the deal, Eye-Fi plans to transition from selling its own brand cards in Europe.

If you’ve ever wished you could clone yourself into two or more people to get more work done, you can be jealous of computers, which can do exactly that. Virtualization allows one physical computer to split itself up into many virtual computers by sharing the computer’s hardware. Chips are now so fast that it makes economic sense to do this, so you can squeeze more work out of each machine. Cloud companies — like, say, Amazon — love it, because it allows them to act a little like a very happy Manhattan real estate developer, and subdivide and rent out a single computer many times over.

Until now, Fusion-io flash memory technology has worked only in cases in which there was no virtualization going on. In big-iron machines that tend to be used for one intensive application at a time, add-in cards are put in servers in order to put data that the process is working on closer to the processor — thus preventing the processor from waiting around, impatiently tapping its foot, for the poky little hard drive that just can’t deliver the data fast enough.

Fusion-io will today announce — at the VMWorld conference in Las Vegas, put on by the virtualization outfit VMware — its ioCache bundle, which is built specifically for virtualized computing environments. Which is pretty much any cloud computing service you’ve ever heard of.

I talked with Fusion-io CEO David Flynn last week, and he told me that the addition of flash speeds gives the physical machine the ability to run as many as three to five times more virtual machines. The benefit, of course, is that you get more work done on a single machine. More work per machine means either higher productivity overall, or savings on the hardware budget — both of which help CIOs score points with the boss.

The ioCache product was created in cooperation with IO Turbine, a company that Fusion-io acquired for $95 million earlier this month.

The company has thus far seen its shares waggle all over the map since its IPO on the New York Stock Exchange in June. Having debuted at $25 a share that day, its stock has traded as high as $36.98 and as low as $19.28. Today, Fusion-io shares closed up $1.05, or more than four percent, to $23.32.

While for tax reasons it’s probably more likely that HP will spin the unit out as an independent company — there are no taxes when assets are distributed to shareholders — Sterne Agee analyst Shaw Wu, in a note to clients issued this morning, breaks down the possible suitors should HP opt instead for a sale.

Topping the list is Samsung, which you might have guessed already. Samsung would make sense, Wu argues, given its “large size and global ambitions.” Samsung has been trying to build a PC business since 1997, when it acquired AST Research, but hasn’t gotten anywhere. But it is the world’s biggest manufacturer of DRAM memory chips, used in PCs; and the largest supplier of NAND flash memory, which forms the basis of solid-state drives that are increasingly built into notebook PCs. It’s also a big maker of LCD displays and notebook batteries. All that vertical integration, combined with HP’s consumer PC footprint — it’s the biggest supplier to Best Buy — would make Samsung the worldwide player it has always aspired to be.

Don’t ignore the other players, though. Acer, Lenovo, Sony and even Dell could all conceivably show up with a bid, Wu writes. But it will all come down to HP’s asking price, and what parts of the business are included. Wu pegs HP’s PC business as being worth $8 billion, or about $3.66 per HP share. To calculate that valuation, he assumes a premium of five times profit of $1.6 billion on $40 billion in revenue; a five percent operating margin and a 22 percent tax rate.

One potential issue to watch in a possible Samsung bid: Whether the South Korean giant asks HP to include its webOS software. Samsung is also a huge supplier of smartphones around the world, and would probably like to rely less on Google’s Android than it does now — and would want to own its own operating system. Having decided to kill the webOS hardware business, HP has indicated that it has plans to keep the software alive in some form, though enough cash from Samsung might change HP’s mind.

Wu also argues that the market has gotten too negative on the PC business in general. While it’s true that Apple’s iPad has left a historically significant mark on the PC universe, PCs aren’t dead yet — just ask Intel. Give them iPad-like touchscreens and flash drives for instant-on capability, and the market might rebound, he says. “We believe longer-term tablets and PCs are the same market. Ironically, we view Apple’s MacBook Air as the first generation of these future hybrid PCs.”

Update at 9:38 AM PDT / 12:38 PM EDT: Samsung just issued a brief statement saying it’s not interested in HP’s PC business. Such rumors are “not true,” the company says. Well it’s really not a rumor exactly, but speculation really. Somehow I don’t believe it’s the last word on the subject.

(Image, obviously, is from the sheet music of the number “Who Will Buy?” from the musical “Oliver!” Hear it below.)

Next we saw Violin Memory — which makes flash-based storage arrays that are intended to make enterprise applications run faster — land $40 million in venture capital funding.

Now we see a third player entering the “flash madness” narrative. Pure Storage is coming out of stealth today, announcing its plans to sell flash-based storage arrays. It is also announcing that it has landed a $30 million C-round led by Redpoint Ventures, with Samsung Venture Investment joining. (Yes, that would be the venture capital arm of the South Korean electronics giant that happens to be the world’s biggest manufacturer of flash memory.) Greylock Partners and Sutter Hill Ventures also participated. The latest round brings Pure’s total funding raised to date to $55 million.

So what is Pure Storage all about? I met up with CEO Scott Dietzen last week and got the download.

The fundamental problem with enterprise storage is that hard drives just can’t keep up with everything else that’s gotten faster in the data center. Flash memory is fundamentally faster, it uses less energy and it takes up less space. We all know this.

The problem with flash is that it has always tended to be more expensive than hard drives. Today, you can buy a one terabyte hard drive for $100 or less. But just try getting that same amount in flash memory and see if the price isn’t, well, a lot higher.

The same principles apply in the data center. CIOs would love to convert to flash-based systems, as long as they’re reliable and affordable and work with the applications and other hardware they already have.

Pure Storage is essentially promising to deliver just that, Dietzen says. The company’s first product is an all-flash storage array that is 10 times faster and 10 times smaller than hard-disk-based systems. It’s called the Pure Storage FlashArray, and it is being aimed at mainstream enterprises in a manner that’s easy to deploy.

Pure’s founders are John Colgrove — one of the founding engineers at Veritas, now part of Symantec — and John Hayes, a founding engineer at Bix, which was ultimately swallowed up by Yahoo. Dietzen hails from Yahoo as well, by way of its acquisition of Zimbra, where he was CTO.

An early key hire was Michael Cornwell, who was lead technologist for flash at Sun Microsystems (now part of Oracle). Cornwell also worked at Apple, where he was Manager of Storage Engineering for the iPod division, and oversaw that product’s transition to — you guessed it — flash memory. Remember the first iPod nano? That was his baby.

Another key name: Greylock venture partner Frank Slootman, the former CEO of Data Domain, is on Pure’s board.

So what’s so special about a storage array built on flash memory? “Disks get slower every year,” Dietzen says. “Intel says processors have gotten 175 times faster over the last 15 years.” Disks just keep getting more data packed onto them, which doesn’t really make them any faster. The mechanical arm inside the disk that grabs data from the platter really can’t go much faster. “Disks today are comparably slower than tape was 15 years ago,” he says.

This creates a problem. Storage needs are going up, but hard drives are slowing data centers down, preventing them from reaching their full potential. It’s only because of cost — about $5 per gigabyte — that hard drives are still appealing. Enterprise-grade flash, on the other hand, tends to cost $40 to $100 per gigabyte, and because flash is historically less reliable, you have to buy double what you really need.

Pure’s play is to get over the cost hurdle. Dietzen says the company can get the cost down to $5 per gigabyte and less.

How does it do that? By reducing the amount of data you actually store. What happens in enterprise environments is that various bits of data get copied and recopied, over and over. Imagine a big filing cabinet with 50 copies of each document scattered around in different folders, when you really only need one. Suddenly the size of that file cabinet need not be so big. The same applies in data storage: Why bother having 10 copies of the same block of data, when one or two will do?

Using a technique known as deduplication, a system can eliminate all those unneeded copies and thus streamline the whole operation. Deduplication, combined with compression, was the primary principle behind Slootman’s Data Domain, which is now part of EMC.

But deduplication is expensive on hard drives, and really doesn’t make sense. Because the mechanical arm in a hard drive is always searching around for where its next needed block of data is to be found, if you employ deduplication, you end up with a bunch of reference signs telling the arm where to go, Dietzen says. The end result is that the disk has to spin more, not less. Flash memory chips don’t have that problem. “We make that process fast, because there’s no performance hit to the deduping process,” he says.

On top of that, Pure has created some algorithms that make the process a lot more granular than on hard-disk-based systems, by working with smaller disk-sector sizes. How small? He wouldn’t say exactly.

Unlike other storage companies — like, say, EMC — Pure’s array, Dietzen says, is built from the ground up for running flash. “The disk-centric companies are slotting flash into places where disks used to be, but they’re not changing the software to take advantage of the flash, to protect the flash from uneven wear and other things.”

A few early companies have tried the hardware, among them the law firm of Fenwick & West, whose CIO Matt Kesner is quoted in Pure’s press release as saying that the data used for various workloads was reduced from 50 to 90 percent.

One key thing that’s going on in the data center these days is virtualization — running several virtual computers within one single physical computer. When you run a lot of virtual machines, you have a lot of data that, like the paper in that big file cabinet, is essentially the same. Dietzen says that Pure’s flash array is able to eliminate a lot of that data. “Even if those virtual machines are a mix of Windows and Linux, there are a lot of commonalities between them,” he says. It’s not uncommon to see the data footprint for virtual machines reduced by a factor of 15 or 20 to one.

And that has caused some interesting reactions among early customers trying out the array. “Some people try it and are shocked when they put 15 terabytes on it and see there’s only one terabyte and think we’ve lost a lot of their data,” Dietzen says. “It’s a little scary at first, but then they run all their workloads and see all the data is there.”

That was around that time of the initial public offering of Fusion-io, the Utah-based start-up that speeds up servers and storage networks. Having opened trading at $25.30 a share on June 9, its first day of trading, its share price has held steady since, and it closed Tuesday at $28.35. It will report quarterly earnings for the first time as a public company on Thursday.

The summer is proving equally interesting for Violin Memory, another company with flash memory based technology that is intended to replace the traditional hard drive based storage arrays that allow enterprise applications like those made by Oracle to run fast. Having raised a $40 million Series C funding round from Toshiba and Juniper Networks at an implied valuation of $440 million in June, the company has been bulking up its staff.

Today Violin will announce that it has named Jonathan Goldick — the former CTO of OnStor, now a unit of chipmaker LSI — as its CTO of Software. Goldick has been knocking around the computing industry for about two decades as an expert on file systems and storage, and his resume includes stints at IBM and Microsoft.

So what does it mean to be CTO of Software at a chip company? Goldick’s job will focus on solving problems related to data management that go beyond the speeding-up that Violin’s technology offers. Once hard drives (which, for all the progress they’ve made in five decades, are still essentially platters of glass; even when spinning at the speed of sound, they are subject to errors and inefficiencies that make them still too slow for the fastest computers) are out of the picture, new problems arise.

“The early adopters, they care about speed because they’re in application hell. But once you get past that, the problem becomes one of data management,” Goldick told me. “Once you make anything 100 times faster or cheaper, you have to revisit how you manage data.”

It’s a big enough problem that Goldick was being heavily recruited by other companies working on bringing flash technology to their own hardware. Goldick wouldn’t name the companies directly, but the hints he dropped suggest he turned down offers from both EMC and Oracle.

Goldick is Violin’s second recent hire. Last month it quietly hired Garry Veale, a former vice president at Hewlett-Packard’s StorageWorks division, as its new managing director for the EMEA region.

There’s a reason that Violin is bulking up its team: The opportunity is potentially huge. Remember, if you will, the December day that Oracle CEO declared that its SPARC T3-4 Supercluster had achieved something of a land speed record of more than 30 million transactions per minute. This was the same speech in which Ellison, in one of his numerous bits of trash-talking, likened HP’s competing product to a turtle. It’s often called “the turtle speech.”

That speech got Violin CEO Don Basile all excited. One of the things that made that Oracle machine so fast was that it was packed with a couple hundred terabytes worth of flash memory. As Basile told me last week: “We loved that speech because they proved us right. It was a big validation for what we want to do.” It also means there’s no end in sight to the flash madness.

The Chromebook line is Google’s first big hardware bet on its cloud-centric Chrome operating system, which is essentially a Web browser capable of running applications that are hosted in the cloud. The point of doing that is that it takes the hassle and the cost of maintaining the software out of the hands of the person or business using it. Google thinks that over the long term it has a chance to erode Microsoft’s dominance of enterprise notebooks.

So what goes into building one? The teardown experts at IHS iSuppli took a look at the 3G version, and have shared their findings with AllThingsD. I talked with analyst Wayne Lam, who worked on the teardown. While you might expect it to essentially be a stripped-down build with as little emphasis as possible on the hardware, it’s actually an interesting study in what can happen when there’s no payment to Microsoft built into the cost assumptions, Lam told me. Added up, the components used cost a total of $322.12.

Since the hardware requirements for storage and memory are lighter, Samsung was able to spend more on hardware that improves the user experience like the display, the battery and the outer enclosure. “The lower overhead in hardware allowed Samsung more leeway on things that people will notice, like a bigger screen and a bigger battery,” Lam said.

The most significant batch of component costs is found on the motherboard (pictured top and bottom, click the images to make them bigger), amounting to $86.37 or about 26 percent of the overall hardware cost. The Chromebook’s microprocessor is a dual-core Intel Atom N570. Samsung, which is the world’s largest manufacturer of memory, supplied its own DRAM chips. Also on the motherboard are power-management chips from Texas Instruments and Intersil.

German chipmaker Infineon supplied a Trusted Computing Platform chip, which is interesting because this is something usually seen in enterprise-level servers and not personal notebooks. The chip helps protect the system by running a thorough security check every time the system is booted up, ensuring that the hardware hasn’t been tampered with and that unwanted software hasn’t been added. Given the bet that Google has made on the cloud, and the attacks it has been fending off of late, this is an understandable move.

Samsung also used its own shop to supply the display. It measures 12.1-inches diagonally and features an improved light-emitting diode technology that boosts its overall brightness. Lam says the display cost $58 and is the second most expensive component in the Chromebook.

The third most expensive component is the battery, which Samsung supplied as well. A key part of the Chromebook experience is long battery life. Samsung opted for a six-cell battery pack that is intended to last all day. It added $48.20 to the hardware cost.

Wireless chips combined for the fourth most expensive set of costs. Hon Hai Precision Technology, the Taiwanese contract manufacturer that’s better known to the world as Foxconn, built the 3G wireless module using four chips from the wireless chipmaker Qualcomm. In order to keep costs down, Samsung opted to use an older Gobi 2000 baseband chip. Wireless chips added $42.85 to the hardware cost. Qualcomm’s newly acquired Atheros unit supplied a Wi-Fi chip.

A few other interesting points that Lam found during the teardown. The memory chips are soldered on to the motherboard, meaning that the computer’s memory isn’t upgradeable by the user as it is on most PCs. It ships with 2GB of memory on board; if you’re a business or educational user paying a monthly subscription fee, by the time you start thinking you may need more memory, it will probably be ready for a hardware refresh. Google is taking even that level of routine management — plus the associated cost — out of your hands. Is the world ready for that?

This morning I ventured down to lower Manhattan to witness the Utah-based company do something its CEO David Flynn swore to me in December that he would not be doing in 2011. But? Things change. After pricing at $19 a share yesterday, the shares opened at $25.30.

Fusion did all the usual things you’d expect from a company wanting to make a splash with an IPO. A big banner with the Fusion-io logo adorned the outside of the NYSE building. And Steve Wozniak, the Apple co-founder who is a Fusion-io investor, director and chief scientist was on hand for the obligatory bell-ringing ceremony. After the bell, eager traders crowded around the Barclay Bank post to await the opening of trading, which took place a little after 10 AM New York time. Woz purchased the first 100 shares when trading opened.

At noon Eastern, Fusion shares are holding up, trading at $23.92. As trading debuts go, it’s certainly no LinkedIn, but then again, on the day LinkedIn opened the Dow Jones Industrial Average closed at 12,605.32, nearly 600 points higher than its open today. Regardless, even at that price, Fusion-io is being valued at nearly $2 billion, or more than 50 times its fiscal 2010 revenue.

Flynn’s next big task will be reporting quarterly earnings next month. Expectations for this company are high, so there’s a lot to worry about. Facebook and Apple combine for about 70 percent of sales but once they’re done building their data centers, they’ll more or less be done buying Fusion-io cards for their servers. And worse, 10 customers account for more than 90 percent of sales.

The good news is that the company has 1,500 different end-user customers. Plus, it has its partners — Hewlett-Packard, Dell, IBM and SuperMicro — to resell its technology into their servers, and there’s lots of interest among financial institutions and other companies in speeding up the flow of data on their servers. It will be an interesting story to watch.

Just moments after the debut in trading, I caught up with CEO David Flynn for a quick chat in the video below.

]]> http://allthingsd.com/20110609/fusion-io-opens-at-25-a-share-worth-nearly-2-billion/feed/ 0 http://allthingsd.com/20110608/flash-madness-continues-fusion-io-prices-at-19-a-share/ http://allthingsd.com/20110608/flash-madness-continues-fusion-io-prices-at-19-a-share/#comments Wed, 08 Jun 2011 23:45:33 +0000 Arik Hesseldahl http://allthingsd.com/?p=84592 Fusion-io, the Utah-based startup that uses flash memory to make ordinary servers run faster just priced for its debut of trading at $19 a share, or six dollars higher than the low end of the original range it gave in its revised S1 filing last month. It’s also higher higher than the $16 to $18 range the company gave in a revised S1 filing yesterday. This should push its opening valuation just shy of $1.5 billion. Read the press release here.

And the shares may open even higher yet. The bid on Fusion-io shares is $29 and the ask is north of $32 according to a couple sources of brokerage data I’ve looked at in the last few minutes. I’m certainly no expert on the mechanics of IPOs, but I think that’s a pretty good indication that the opening trades will be a lot higher than $19 a share.

As previously noted, Fusion’s biggest customers include Facebook and Apple, which both use the technology in their data centers. Others include Hewlett-Packard, IBM, Dell and Supermicro, all of which resell servers using Fusion-io’s technology.

Goldman Sachs and Morgan Stanley are acting as lead active joint book-runners for the offering while J.P. Morgan Securities and Credit Suisse are acting as passive joint book-runners. The company will trade on the New York Stock Exchange under the symbol FIO. I’ll be down at the Exchange tomorrow morning to witness the start of trading. I’ve never been to an IPO before.

At $19 a share, the holdings of CEO David Flynn holdings will be worth north of $130 million. He’s putting up 500,000 shares for sale, which at that price would be worth $9.5 million. After the offering he will control a little less than 8 percent of the equity in Fusion-io. Chief Marketing Officer Rick White owns nearly 5.7 million shares, which will at that price be worth more than $100 million.

The venture capitalists who backed it will do very well New Enterprise Associates owns nearly 26 million shares which at $19 a share would be worth nearly half a billion dollars. Lightspeed Ventures owns another 8.8 million shares, which would be worth more than $167 million.

]]> http://allthingsd.com/20110608/flash-madness-continues-fusion-io-prices-at-19-a-share/feed/ 0 http://allthingsd.com/20110607/flash-madness-fusion-io-ipos-thursday-but-first-violin-raises-40m/ http://allthingsd.com/20110607/flash-madness-fusion-io-ipos-thursday-but-first-violin-raises-40m/#comments Tue, 07 Jun 2011 12:45:05 +0000 Arik Hesseldahl http://allthingsd.com/?p=83415 This week is going to be a big one for companies in the business of bringing flash memory chips to the data center. The main event will be the Thursday IPO debut on the New York Stock Exchange of Fusion-io, a company I’ve written about here numerous times.

However, as a warmup, another flash company, Violin Memory, announced today that it has closed a $40 million Series C round of funding at an implied valuation of $440 million. Violin, based in Mountain View, Calif., is run, oddly enough, by Don Basile, a former chairman and CEO of Fusion-io. Obviously, he will be watching that company’s opening days of trading with significant interest, presumably because he still has some equity in it, but also because of the implications for his new company, which he’d like to take public as well.

Where Fusion-io sells flash-based cards that make servers run faster–Facebook and Apple buy them for use in the servers running inside their data centers and between them constitute about 70 percent of its revenue–Violin sells flash-based memory arrays that are intended to replace the hard disk-based memory arrays that make enterprise applications run faster. Violin’s arrays come in a range of sizes from tens of terabytes up to hundreds of petabytes, and are said to significantly speed up Oracle and other databases by a factor ranging from 10 to 100 depending on the situation.

HP has set speed records running Microsoft’s SQL Server using Violin arrays, Basile told me. Not only does it speed them up, but the Violin arrays eliminate 80 percent of the required hardware footprint and reduce the necessary power by 90 percent, cutting back on operational costs. Hewlett-Packard resells Violin arrays, and AOL is a big customer, Basile told me.

Violin hasn’t been raising money via the traditional venture capital route. Its investors have included Toshiba, the Japanese electronics concern that happens to be a big manufacturer of flash memory chips used in the arrays, and Juniper Networks. It has also taken money from large funds that dabble in private investments, and from several wealthy individuals, among them Atiq Raza, the former number two at Advanced Micro Devices, Arjun Gupta, the founder of TeleSoft Partners, and venture capitalist Dixon Doll. Basile has taken investments from nine such individuals, and these are only three that he named. He also stressed that they are personal investments.

Violin raised $35 million earlier this year in a Series B, and raised $10 million in a series A last year. In addition, the company has $140 million in combined debt and credit, giving it a combined $180 million to fund its operations and growth for the foreseeable future. The company expects to do more than $100 million in revenue this year.

Still, Basile would like to go public, and will be watching the IPO both of Fusion-io and of Groupon to see how the market reacts to them. If they react well, he says he plans to hire bankers by the end of the summer. “I don’t have a compelling reason yet, but if the markets react favorably it would be in our interest to look at the public option very seriously,” he said.

Meanwhile, Fusion-io will debut under the trading symbol FIO on Thursday. You can expect CEO David Flynn to make the rounds with a series of interviews tomorrow on CNBC and elsewhere. When last heard from, the company said in an updated S1 filing with the U.S. Securities and Exchange Commission that it expected to price in the range of $13 to $15$16 to $18 a share, in order to raise $185 million. A price in that range would value the Utah-based company north of $1 billion $1.4 billion (see today’s latest updated S1 filing here). It is one of two companies set to go public on Thursday, the other being Taomee Holdings, a China-based company that produces media for children.

As Dow Jones Newswires noted yesterday, Fusion-io’s debut is coming against the backdrop of a market that has been declining in recent weeks, giving it a certain headwind. And there’s already been plenty of criticism of Fusion-io’s prospects. As noted, two customers, Facebook and Apple, account for about 70 percent of revenue, while 10 customers account for more than 91 percent of revenue. (The Wall Street Journal commented on this in March.) As risks go, a concentrated set of customers is a classic one. If one or two suddenly stop buying–a fair risk when you consider that Facebook and Apple will soon complete construction of their respective data centers–then sales can drop just as suddenly. Investors like seeing a large, diverse customer base.

I asked Basile about that and whether the same long-term risk applies to Violin. “It is a legitimate risk,” he says. Violin doesn’t have the same kind of concentration. While AOL is a big customer, he says, Violin has no significant customer who accounts for more than 50 percent of sales. Its revenue splits roughly even, with about half of its sales coming from “channel” customers who build Violin’s memory arrays into their own products, while the other half buy Violin products directly to integrate with systems they’ve purchased from other vendors.

Then he pointed to language in Fusion-io’s S1 filing as a way of making a point about the wider prospects for flash memory use in the data center. Yes, Fusion-io has a handful of big customers, but it also has more than 1,500 end-customers. Among those are customers of Hewlett-Packard, IBM, Dell and Supermicro, who all sell Fusion-io’s cards as an option on their own servers. Among them they sell about 9 million servers a year, and if you do the math, he says the current run rate reveals that 12,000 of those servers have flash cards from Fusion-io. “That leaves a lot of room for growth.” Room enough for both companies, he said.

]]> http://allthingsd.com/20110607/flash-madness-fusion-io-ipos-thursday-but-first-violin-raises-40m/feed/ 0 http://allthingsd.com/20110523/at-13-to-15-a-share-fusion-io-will-be-worth-more-than-1-billion/ http://allthingsd.com/20110523/at-13-to-15-a-share-fusion-io-will-be-worth-more-than-1-billion/#comments Tue, 24 May 2011 00:45:00 +0000 Arik Hesseldahl http://allthingsd.com/?p=77099 Fusion-IO, the enterprise storage company that uses flash memory to speed up conventional servers, updated the paperwork on its forthcoming initial public offering today. It disclosed that it plans to sell its shares in the price range of $13 to $15 which would give it a valuation just shy of $1.2 billion and would, the company hopes, raise about $212 million in the process.

That’s a considerable increase over the $150 million it initially sought when it filed its first S1 with the U.S. Securities and Exchange Commission in March. Fusion-IO surprised a lot of people with that filing, mainly because it had been so successful raising money from venture capitalists. As recently as December of last year, CEO David Flynn had dismissed the idea that Fusion-IO would go public in 2011.

A few more interesting details from the new filing: It reiterates the disclosure that Facebook is a significant customer and adds Apple to that list. The top three during the 2010 fiscal year were IBM, which resells Fusion-IO products in its own servers, which accounts for 13 percent of sales; Hewlett-Packard, also a reseller at 10 percent; and Facebook, which buys directly at 10 percent.

That changed significantly in 2011: Facebook accounted for 52 percent of sales and Apple for another 20 percent during the first nine months of fiscal 2011, meaning its top two customers account for 72 percent of sales. It also says its ten largest customers amounted to 91 percent of its revenues in the first nine months of the year, up from 75 percent in 2010 and 47 percent in 2009. It further warns that sales to Facebook will decline significantly in the quarter ending June 30, presumably because Facebook will be finished building out its data center. It notably says nothing of the kind about Apple.

Sales were $125 million in the first nine months of the fiscal year, and the company ran a $1.2 million loss amounting to 9 cents a share. It did, however, report a profit in its most recent quarter, its first, of $7 million on $67.2 million in revenue, for the quarter ended March 30.

Assuming the stock prices at $15 a share, Flynn’s holdings will be worth north of $100 million, though he’s putting up 500,000 shares for sale, which at that price would be worth $7.5 million. After the offering he would own a little less than 8 percent of the company.

Other big winners will be the venture capital investors who backed the company. New Enterprise Associates controls a little more than 38 percent of the equity in Fusion-IO, and owns nearly 26 million shares which at $15 a share would be worth nearly $390 million. Lightspeed Ventures owns another 8.8 million shares, which would be worth more than $132 million.

]]> http://allthingsd.com/20110523/at-13-to-15-a-share-fusion-io-will-be-worth-more-than-1-billion/feed/ 0 http://allthingsd.com/20110328/the-nintendo-3ds-appears-pretty-profitable-judging-by-the-teardown/ http://allthingsd.com/20110328/the-nintendo-3ds-appears-pretty-profitable-judging-by-the-teardown/#comments Mon, 28 Mar 2011 13:30:05 +0000 Arik Hesseldahl http://newenterprise.allthingsd.com/?p=4430 Nintendo’s latest handheld gaming device has hit the market in Europe and North America and as so often happens, before the weekend was over my in-box contained a detailed teardown report from the team at IHS iSuppli.

As usual, the idea behind the teardown is not only to figure out who Nintendo’s component suppliers are and what parts are being used, but to estimate how much all the components cost to help guess how much of a profit margin Nintendo is making on each unit. And it looks like a decent margin. ISuppli says the cost of all the parts in the device itself plus what’s in the box amount to $103.25 for a device that’s selling at retail for $249. The cost works out to an increase of about $25 over the Nintendo DSi, the most recent Nintendo handheld, released in 2009, which cost about $78, when iSuppli tore it apart that year.

While most of the components come from Japan, it’s not entirely clear if the supply of any of the parts used come from areas affected by the earthquake and tsunami, says Andrew Rassweiler, an iSuppli analyst who supervised the teardown. “Many of these component should have a greater risk exposure to supply chain problems, though we don’t know about any specific disruptions at this point,” he said.

The most expensive component, as is often the case with consumer electronics, is the displays. The 3DS uses two Sharp displays that cost a combined $33.80. The headliner is the top screen 3D. It’s a 3.5-inch 800-by-240 pixel display that uses an LCD-based parallax barrier panel sandwiched to the back of the color LCD which alternates between the left and right images at a high rate of speed to produce the 3D effect. “It looks like a conventional LCD from the outside, but when you open the display you see that on one side of the glass is essentially the conventional color element, and on the other side of the glass is a monochrome element,” Rassweiler told me. “It’s a clever bit of display engineering.”

The handheld’s main chip is an applications processor. It’s a custom ARM-based chip manufactured by Sharp, that at a cost of $10.02 is only slightly more expensive than the chip in the previous Nintendo DSi. However, Nintendo has quadrupled the amount of flash memory in the 3DS versus the DSi to 16 gigabytes, and Samsung, the world’s largest manufacturer of flash, supplied it. Fujitsu supplied another type of memory known as fast-cycle RAM. Rassweiler says for this particular type of memory, Nintendo has used a type of chip that’s only made by Fujitsu, which is odd because FCRAM is widely available, and its unusual for consumer electronics manufacturers to “single source”–that is, rely upon a single supplier for an important component. The combined cost of memory on the 3DS worked out to $8.36, more than twice the cost of the memory found on the DSi.

Three chips related to the user interface cost a combined $6.81: an accelerometer from STMicroelectroncis, a gyroscope from Invensense, and an audio chip from Texas Instruments. Atheros, the Wi-Fi chipmaker that’s being acquired by Qualcomm, supplied a $5 Wi-Fi chip. TI and NEC supplied power management chips that cost $3.63. The 3DS contains three cameras, and though it’s not clear who supplied them–camera suppliers have gone to great lengths to hide their identities in recent years–iSuppli reckons their combined cost at $4.70.

Since I often get asked this question, let me say that iSuppli’s analysis focuses strictly on the materials used and doesn’t account for the cost to develop software or to license any patents. Nor does it account for the cost of any shipping or distribution or marketing. It’s just the raw cost of the hardware.

]]> http://allthingsd.com/20110328/the-nintendo-3ds-appears-pretty-profitable-judging-by-the-teardown/feed/ 1 http://allthingsd.com/20110321/japans-quake-cuts-into-supplies-of-raw-materials-used-in-chips/ http://allthingsd.com/20110321/japans-quake-cuts-into-supplies-of-raw-materials-used-in-chips/#comments Mon, 21 Mar 2011 17:30:21 +0000 Arik Hesseldahl http://newenterprise.allthingsd.com/?p=4175 After more than than a week of gathering anecdotal reports about shortages here and there, the research firm IHS iSuppli has concluded that 25 percent of the world’s supply of silicon wafers used to make chips has been been suspended by the effects of the earthquake and tsunami in Japan.

Manufacturing has stopped at Shin-Etsu Chemical Co. Ltd.’s Shirakawa facility, and MEMC Electronic Materials has stopped manufacturing at its plant in Utsunomiya. Together, the two facilities account for a quarter of the global supply of silicon wafers, the basis of building chips.

The Shin-Etsu Chemical plant by itself supplies about 20 percent of the world’s silicon supply, and it specializes in making 300-millimeter wafers, which are the dinner-plate-size discs of silicon used in the more advanced chip factories, commonly referred to as fabs. Shin-Etsu, iSuppli says, supplies several memory chip manufacturers, particularly those that make flash memory, used in everything from iPhones to memory cards, and also DRAM, the main memory used in PCs and servers. ISuppli says the global market is going to be hit hard, which in turn means you can expect prices on both flash and DRAM to soar. Shin-Etsu has said it would set up production at other plants, but it’s hard to know how long that will take.

MEMC’s Utsunomiya facility accounts for five percent of worldwide wafer supply. MEMC said it expects that shipments from this facility will be delayed during the near term.

In a related note, iSuppli has quantified the impact of the shutdown of operations at Mitsubishi Gas and of Hitachi Kasei Polymer. The two companies produce about 70 percent of the world’s supply of the raw materials used to make printed circuit boards. The key material in question is called copper-clad laminate or CCL. The two companies say they’ll be able to ramp production back up within two weeks. The good news is that electronics manufacturers have enough circuit boards in inventory that they can probably keep their operations running without interruption.

ISuppli goes on to check in on a few chip companies in the affected region: Elpida Memory says its fab in Yamagata has been damaged, and the lack of electricity is hurting production. It’s running at about half its normal capacity.

The quake also damaged about 40 percent of the production capacity of Renesas Electronics. Production has stopped at its Tsugaru fabs where it makes analog and discrete chips, at its Naka fab where it makes system-on-chip and microcontrollers, and at its Takasaki and Kofu fabs, which also making analog and discrete parts.

Half of Fujitsu’s production capacity has been damaged. While its fabs and wafer equipment are intact, the lack of power, gas and wafers have slowed things down considerably, and it expects to recover in about three to four weeks.

One company that is holding up well: AKM Semiconductor, notable for the compass chips it produces for Apple that are used in the iPhone and iPad 2. Its main production fab in Nobeoka is well out of the quake zone and hasn’t suffered any loss of power.

]]> http://allthingsd.com/20110321/japans-quake-cuts-into-supplies-of-raw-materials-used-in-chips/feed/ 5