Fabric Networks Rips at Seams

InfiniBand startup formerly known as InfiniSwitch shuts down. Is the infini-bloodshed over?

September 11, 2003

3 Min Read
Network Computing logo

InfiniBand startup Fabric Networks Inc. -- formerly known as InfiniSwitch -- shut its doors yesterday and laid off most of its employees after it failed to find traction, CEO and president Alisa Nessler tells Byte and Switch in an exclusive interview.

The Westborough, Mass., firm, formed in March of this year by the merger of Lane15 and InfiniSwitch, changed its name to Fabric Networks in June. It was trying to sell InfiniBand switches for high-speed server-to-server interconnection (see InfiniSwitch Now Fabric Networks and InfiniSwitch Merges Lanes).

Unfortunately, the name change didn't do the company any good.

Nessler says that most of the company's 30 remaining employees have been laid off, with a small team staying on to explore the sale of its assets. "The market is still a tough, tough market," she says. "It was really about timing of the revenue opportunity and the return on investment our investors were looking for."

The two companies that merged to form Fabric Networks had raised a total of $52 million in funding. Lane 15 had secured $22 million from investors including Index Ventures, Quanta Computer Inc., Convergent Investors, AV Labs, Austin Ventures, Dell Computer Corp. (Nasdaq: DELL), Intel Capital, and Lightspeed Venture Partners. [Disclosure: Lightspeed is an investor in Light Reading Inc., publisher of Byte and Switch.]InfiniSwitch, founded in September 2000 and based in Austin, Texas, had received a total of $30 million from investors including Bessemer Venture Partners, OneLiberty Ventures, Newcogen Group Inc., Columbia Capital, TL Ventures, and Moore Capital Management (see InfiniSwitch Bucks Boost InfiniBand).

The demise of Fabric Networks is yet another sad tale for InfiniBand, which has been one of the most disappointing emerging technologies in recent memory. Once hyped as a high-speed, low-latency replacement for multiple computing and network interconnects, InfiniBand has been relegated to the corners of the high-performance computing market. Other InfiniBand companies that have gone belly-up so far include Paceline Systems, which sold its intellectual property to Motorola Inc. (NYSE: MOT), and OmegaBand (see Motorola Acquires Paceline's Assets and OmegaBand Is InfiniBusted).

Meanwhile, other IB startups, including Banderacom Inc. and VIEO Inc., have distanced themselves from the technology after it became clear that InfiniBand would not bloom as quickly or as widely as first thought (see Banderacom Abandons InfiniBand and VIEO Vamps Up $5.5M).

Major technology vendors have also backed away from InfiniBand. Last year IBM Corp. (NYSE: IBM), Intel Corp. (Nasdaq: INTC), and Microsoft Corp. (Nasdaq: MSFT) all cancelled their InfiniBand product plans (see IBM Kills InfiniBand Chip, Intel Bails on InfiniBand, and Microsoft Backs Off InfiniBand).

At the same time, Sun Microsystems Inc. (Nasdaq: SUNW), Oracle Corp. (Nasdaq: ORCL), and IBM's server group remain bullish on InfiniBand's ability to cluster servers together (see Server Vendors Hold IB Pep Rally). Hoping to carve out a piece of this market are surviving InfiniBand switch startups InfiniCon Systems Inc., Topspin Communications Inc., and Voltaire Inc., as well as InfiniBand silicon supplier Mellanox Technologies Ltd. (see Sun, Topspin Team on Infiniband, InfiniCon Shrinks Switches, Voltaire Mates InfiniBand & iSCSI, and Mellanox Ships 100K IB Ports).Krish Ramakrishnan, CEO of Topspin, believes InfiniBand has actually just started to make some serious inroads with high-level support from Oracle and Sun, which is licensing Topspin's technology (see Sun Circles Topspin). "It's ironic that Fabric Networks is shutting down," he says. "The traction we're seeing with InfiniBand right now is tremendous."

Nessler, who founded Lane15 and served as its CEO, says Fabric Networks and its backers still believe in the promise of InfiniBand. The problem, she says, was simply that it was too early.

"There was not enough revenue traction and visibility soon enough," she says.

Todd Spangler, US Editor, Byte and Switch

Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like

More Insights