U.S. startup Fusion-io, backed by computing pioneers Steve Wozniak and Michael Dell, says it can outwit computer-storage incumbents like EMC which have too much invested in old technology.

Fusion-io, which has filed for a U.S. initial public offering, believes its fresh approach to speeding up data retrieval coupled with a first-mover advantage mean it can win deals over rivals concerned with bigger things.

Established companies go after known markets. We created a new market, and that market is for storage that runs at the speed of memory, David Flynn, Fusion-io's chief executive, told Reuters in an interview.

Fusion-io says its technology speeds up data processing by a factor of 10 by putting a solid-state flash memory drive directly into the computer server rather than in the traditional storage area, which is further away and takes longer to reach.

It was a counterculture in disk storage, said Wozniak, who created the Apple I and Apple II computers that helped revolutionize personal computing in the 1970s, and was hired by Fusion-io as its chief scientist in 2009.

Wozniak told Reuters his work at Fusion-io involved not a lot but said he believed in the way the company fused hardware and software, much as he had done personally with the first Apple computers.

Fusion-io has already raised more than $100 million from investors including Dell's CEO and founder Michael Dell, Korean electronics group Samsung and venture capital firm Accel.

It wants to raise up to $150 million in an IPO on the New York Stock Exchange.

Fusion-io has been early to turn its attention to input/output, or I/O -- the transfer of information between a computer's central processing unit and its main memory.

Until recently, most efforts to make computing more efficient have been concentrated on increasing processing power or memory in isolation.

It's what is the limiting factor today, with how much data can convert into knowledge. It is the bottleneck, said Flynn.

Steve Duplessie, senior analyst at IT research firm Enterprise Strategy Group, said Fusion-io's short-term prospects were good but he had reservations about the longer term.

I think they have been very innovative and original in their approach to make a single server go fast, he said.

The negative is that the world of single servers is going to end -- the point of infrastructure is to share everything you have with everything else.

A version of their acceleration technology that can be leveraged/shared between many physical servers (hosting even more virtual servers) is what will win, and I've seen nothing on the Fusion road map to tell me they are working on that.

However, Fusion-io's Flynn said most competitors had neither the technology nor the motivation to catch up fast.

The technology itself is very non-trivial, he said. You have to teach these operating systems how to consume something that is entirely new and foreign to them.

He added that large data-storage companies would find it hard to sell technology that was much more efficient than what they currently sell, since it would eat into their sales.

The people like EMC and NetApp -- even server vendors themselves -- are conflicted, he said. They can't cannibalize their own revenue streams.

(Editing by David Cowell)