Did a ‘poison pill’ threaten Intel's StrongARM licence?
An odd bit of Digital lawyering may have had an unforeseen affect. Allegedly...
Earlier this week Intel and ARM agreed a licence deal (see story), and ARM's share price went crazy. The deal means future development of Intel's StrongARM processors is assured, and the endorsement of Chipzilla is of course mightily important for ARM. But the funny thing is, Intel had an ARM licence already. Exactly why Monday's licensing deal was hugely important while the one announced on 23rd February 1998 isn't is shrouded in some considerable murk, as the terms are confidential. The 1998 announcement said that the licence allowed Intel "to produce, sell, and enhance the StrongARM microprocessor family." It also said that there was a cross-licensing deal involved. This week's announcement says the two "have finalised a licensing agreement which will enable Intel to develop a full range of solutions based on current and future versions of the ARM architecture beginning with version 5TE." The specific mention of the 5TE is one possible clue to what's different this time around. ARM tells us that the previous licence was effectively inherited from Digital, and that it covered ARM4 and SA-1 StrongARM. The latest licence covers ARM5 and SA-2 StrongARM, so although Intel already had a licence, this is the first time it's actually asked for one. That explains (perhaps) the curious quote from ARM worldwide marketing VP Reynette Au, which suggests that Intel just came on board the ARM team this week: "Intel has joined the growing ranks of companies that have embraced the ARM architecture and the ARM Partnership Model," said Au, without referring to the earlier club membership. Mind you, it's strange that Intel started talking about SA-2 in March, but didn't actually secure the licence it needs to build it until October. A lot of wrangling, perhaps? Well, we hear a funny story, which is sourced to a very highly-placed ARM executive. The original StrongARM licence was with Digital, and Digital's lawyers, we're told, wanted to assure the future of the line in the event of ARM being taken over. So they suggested a clause which assigned the architecture entirely to Digital if this happened. ARM agreed, and suggested a quid pro quo - if Digital was taken over, then ARM should get to own StrongARM. Agreed, and what happened next? Well, if there's truth to the above, what happened next is at least a bit debatable. StrongARM was transferred to Intel as part of the settlement arrangements to Digital's lawsuit against the company. But after that started rolling, Compaq bought Digital. You can see how the timing of this could be rather important if the story of the ARM-Digital contra clause is true. Could we consider the Intel-Digital deal a done one prior to the Compaq takeover, even though the takeover threw the whole thing up in the air? Was the Digital StrongARM licence transferable in the first place? To what extent was the 1998 Intel licence a new one, rather than a transfer? How many lawyers and millions of bucks would it take to sort this little lot out, if Intel and ARM couldn't come to a mutually agreed solution? Given the circumstances of the original Intel-Digital deal it's perfectly plausible that the IP could have gone a bit haywire. Intel didn't actually want StrongARM in the first place, but was essentially shipped it on Digital's insistence as part of a big package of agreed measures. Intel then took a while to figure out what to do with StrongARM, at which point it would have started to figure out whether it actually owned it or not. Even if the contra clause story is true the timing of the various signings will have had an effect on this, and matters will have been confused enough for it to be difficult to say with any certainty who owned StrongARM, and what the courts might have said about who owned it. But the thought of Intel's attorneys combing the licence agreements and discovering that Digital had transferred nothing much at all to Intel is one to treasure. ®
Sponsored: Benefits from the lessons learned in HPC