Sun Sets on Azul's Technology

Did startup Azul use Sun Microsystems' trade secrets to develop a hot new server? There's a potentially nasty legal wrangle in the works

In the fall of 2004, Azul Systems Chief Executive Stephen W. DeWitt paid a visit to Sun Microsystems CEO Scott G. McNealy. Azul was weeks away from unveiling a server that would compete with Sun products, and DeWitt was determined to pay his former boss a courtesy call. "I have a lot of respect for him, and for a lot of people at Sun," DeWitt says. "I could have let him read about [Azul's new machine] in BusinessWeek, but that's not what we did. We were open with him."

If DeWitt wanted to remain in Sun's good graces, he has failed miserably. For the past year the companies have been wrangling over Sun's belief that the Azul system was designed with healthy helpings of Sun (SUNW) technology and trade secrets. Faced with what it says was an imminent lawsuit from Sun, Azul made a bold end run.

Azul filed suit against Sun on Mar. 14, seeking a declaratory judgment that it did not infringe on 20 specific Sun patents or misappropriate any Sun trade secrets. "[Sun] made accusations that we knew were false, but [it] never agreed to get into any process where we could get into a common understanding of the truth," DeWitt says. "All [it] did was threaten us over and over and over with litigation."


  Azul's move sets the stage for what could become a nasty spat between two companies with much in common. Both have charismatic leaders, known for high-energy personalities and an uncanny knack for humor-laced comments. Azul's servers are designed to run software built with Sun's Java programming technology. And Azul's 207 employees include a dozen or so Sun alumni, including former Sun marketing chief Shahin Khan.

DeWitt himself worked for Sun, after his first startup, Cobalt Networks, was purchased by Sun for $2 billion in 2000. DeWitt says he never felt ill will from McNealy. Insiders say McNealy disliked the deal from the start, however, and that his distaste only deepened when Sun had to take an embarrassing write-off on Cobalt in 2003.

Sun is not talking about the spat, beyond issuing a harshly worded statement. It reads, in part: "Sun has spent over a year trying to achieve a business resolution to Azul's unauthorized use of Sun [intellectual property]. During this period, Azul has repeatedly stonewalled and delayed."


  Analysts and former Sun executives say they're somewhat surprised by the row, as Sun has filed relatively few intellectual-property suits over the years. Indeed, its current marketing mantra is "share" -- a campaign built on Sun's belief in developing and supporting industry-standard technologies such as Java.

But Azul tells a different story. It says that from the time Sun first raised concerns in February, 2005, Azul has bent over backwards to prove that the fears were unfounded. Dewitt says he called McNealy after receiving the first letter from Sun's legal department. In the ensuing months, Azul had its outside counsel do an "intellectual-property audit," combing through documents and interviewing former Sun employees on staff to make sure they had not wittingly or unwittingly misused Sun patents or trade secrets. An Azul insider says Sun was uninterested in seeing the results.

Azul says in a court filing that it made other proposals, including "a mediated discussion between the parties' technical teams or other form of mediation or arbitration." Sun's response came in a letter dated May 16, 2005: "We do not need a mediator or independent auditor to point out the obvious," according to Azul's filing.


  By February, 2006, the situation was reaching breaking point. On Feb. 16, a Sun executive told an Azul board member that Sun planned a lawsuit, and that he had "seen a draft of the complaint," the filing says. In March, Azul offered to let Sun look at its documents and interview its engineers, if Sun would agree not to sue based on what it found. Sun refused.

Instead, DeWitt says, Sun delivered an ultimatum, demanding tough terms it had raised before: either give Sun a significant equity stake in Azul, and pay what DeWitt calls "exorbitant" licensing fees and royalties for the use of Sun technologies -- or face Sun's legal wrath. One Azul insider says Sun was seeking around 20% of Azul's equity.

At that point, Azul decided to file its suit -- in essence, beating Sun to the courthouse so as to better define the rules of engagement. It's a risky strategy. "Suing the company that owns the patent?" asks one former Sun executive. "I just don't get it. This is like the mouse slapping the cat in the jaw. They're just asking for it."


  One reason for the move is to set the venue in Silicon Valley, home to both companies, rather than take a chance that Sun would file in a plaintiff-friendly locale, such as the U.S. District Court in eastern Texas, says Heller Ehrman attorney Robert Haslam, who is Azul's outside counsel. "No matter what Azul did to explain [its] technology and why it wasn't derivative of Sun's, negotiations weren't going anywhere," says Haslam.

Why is Sun so interested in Azul? The startup is pursuing a potentially game-changing plan that could save customers time and money: hugely powerful "network-attach processing" systems that corporations could use to run any Java-style program -- eliminating the need to parcel out applications among reams of less-powerful devices, or to hire the staff and buy the "grid-computing" software to tie them all together.

The idea came to two of Azul's founders, Shyam Pillalamari and Gil Tene, during the JavaOne trade show in 2001. As they walked the show floor, Tene called Pillalamari via cell phone to share his insight: that gobs of software developers were no longer even questioning whether future products would be built with Java. So why not start a company to create a computer that specialized in running such code, without having to waste any energy running the hulking operating systems inside conventional computers -- such as Microsoft's (MSFT) Windows, Sun's Solaris, or the open-source Linux program -- inside the server.


  At that time, there was little Sun connection within Azul. That changed once DeWitt joined as CEO in October, 2002. Khan joined two years later.

Even as the negotiations began turning ugly in 2005, Sun executives publicly dismissed Azul as inconsequential and doomed for failure. But the startup was making impressive progress in some regards.

Azul completed development of its radical new "Vega" chip architecture, which stuffs as many as 24 separate "cores" onto one silicon, or as many as 384 when clustered together. It has been shipping products for two quarters, an accomplishment in itself in an industry where there have been few totally new designs in decades.

More recently, Azul has signed a few deals with longtime Sun customers, such as Pegasus Solutions and Credit Suisse First Boston (CSR). In fact, CSFB was an investor in a recent round of venture funding that raised more than $30 million, says an insider.


  "Now that we're getting some traction, it's very clear that Sun is threatened by us," says DeWitt. "We announce some customer wins, and all of a sudden they go from perpetual threats to drawing a line in the sand, where they say, either you acquiesce to our demands, or we'll sue."

Many observers say Sun will indeed file the lawsuit. That means hefty legal expenses for Azul, and one more reason for would-be customers to forgo taking a chance on a startup. That's got the normally affable DeWitt steaming. "Getting involved in a lawsuit is a heroic waste of time and a heroic waste of money," he says. "But we have to defend ourselves from a much larger predator." Just how well Azul fares is likely to be discovered in court.

Before it's here, it's on the Bloomberg Terminal.