Over the last few quarters, AMD has made it clear that it’s looking for new sources of revenue, including potential licensing deals and IP arrangements. Now Bloomberg is reporting that Intel is one potential company interested in licensing some of AMD’s patents, and that the two companies are in talks about a potential deal.
Presently, Intel has a patent licensing deal with Nvidia with a total value of $ 1.5 billion and an expiration date of March 17, 2017. That deal has earned Nvidia roughly $ 67 million per quarter, and while that’s not much compared to the company’s yearly revenue of $ 5 billion for fiscal year 2016 (calendar year 2015), it works out to about 5% of its revenue and a significant chunk of Nvidia’s $ 614 million net income. Companies like IP and patent licensing precisely because these multi-year agreements can bring in continuing funds for work already completed without being a further drain on the bottom line. Details on the current state of negotiation between AMD and Intel are unknown, as are dollar figures and licensing terms.
There’s already been some signs that Intel and Nvidia might not renew that agreement, including Intel’s stated intention to adopt the VESA Adaptive Sync / FreeSync technology rather than paying for a license for Nvidia G-Sync. We’ve also seen signs of limited cooperation with Intel via the recently-announced XConnect external GPU hardware. It’s possible that AMD is willing to be more flexible on patent terms and arrangements than Team Green — AMD’s financial situation isn’t great, and while the company expects new revenue in 2016 from a third semicustom embedded design win (currently hypothesized to be the Nintendo NX), the company needs all the revenue it can muster to offset lower APU sales. Successful IP licensing agreements, even if they weren’t for very much money in absolute terms, would also help CEO Lisa Su argue that she’s successfully pivoting the company in new directions to take advantage of its existing IP and patent portfolios.
Intel’s multi-year patent deal with Nvidia hasn’t produced any GeForce-branded integrated graphics hardware, and it’s unlikely that Intel would explicitly adopt AMD’s architecture. Instead, the license would cover lower-level technology and implementations that Intel could incorporate into its own integrated graphics, or possibly Xeon Phi if AMD has data center or HPC-related patents that Santa Clara wants to take advantage of. AMD might also have patents related to HBM that are of interest to Intel — while the firm hasn’t announced any plans to utilize that technology, it could be interested in long-term applications or in using low-level design elements for other tasks.