Intel – What Else Will They be Inside?
A lot was made of Intel’s earnings last week, and for good reason. Anyone that’s followed the company knows that a PC recovery is an extremely accurate signal for renewed consumer spending and for the overall direction of the economy. Needless to say they had the biggest y/o/y revenue increase in the company’s history.
Along with the earnings came a slew of predictions about how Intel will make out as they continue to chase down the mobile market – even they know that PC growth won’t track well into the next decade… they failed in the mobile sector in the past and sold off their X-Scale assets to Marvell several years ago. But Intel is never one to give up easily.
A friend asked me what I thought about this article: Analyst: Intel may acquire FPGA vendor, which discussed the potential for an acquisition of Altera or Xilinx, the kings of the FPGA – or programmable logic – market.
Acting like an investor or a corporate development guy, I pounded away my rough thoughts off to the top of my head, MBA hat on, on whether this marriage would make any sense.
1. Legal / Strategic: part of Intel’s success has come from its monopolistic behavior and anti-competitive business practices (the recent AMD settlement was a glaring admission of wrongdoing by Intel / Otellini).How would an FPGA acquisition be viewed by government? There have been reoccurring acquisition rumors over the past few years of Intel buying nVidia, and there appeared to be hesitancy by the analyst community on whether this would pass scrutiny, since it would take a market (graphics) with two entrenched competitors (nVidia vs ATI [now AMD]) and allow Intel to potentially invest and squeeze out the competition (anti-competitive pricing etc). Any move into FPGAs by Intel would need to be discussed and analyzed in the context of whether it would draw anti-trust scrutiny. We’re unlikely to see a repeat of Intel’s past practices in the communications IC sector, since Intel itself will be reluctant to use strong-arm tactics that allowed it to gain dominance in PCs.
Instead, Intel may be more likely to purchase smaller companies and startups who are designing innovative technologies that will prevail in the embedded space. As a comparison, Google appears to have recently moved to this strategy, since large acquisitions really signal an intent to dominate and “roll-up” specific markets, which raises anti-trust concerns and draws consumer ire. Intel may therefore be more likely to make smaller technology acquisitions using its massive free cash flows, which will give the appearance of technologies being designed “inside Intel”.
2. Synergies: with Intel’s recent focus on embedded computing and their goals for the mobile segment, Intel is signaling a desire to provide single-chip subsystems incorporating complex logic and multiple processor cores. One main advantage I see for Intel in having an established FPGA product line is to provide “whole solutions” in embedded. Their purchase of WRS signals this intent. For example, in the server market, Intel is well known for “bundling” chipsets with CPUs and providing these entire solutions to OEMs. E.g. they can throw in a PCI-Express or north bridge chip for free to a customer buying the CPU. This capability allows Intel to undercut competitors and maintain high margins on its core businesses (CPUs).
Intel is undoubtedly directing WRS to work on the difficult issues around software for multi-core / parallel compute environments, since this is the main trend / challenge in CPUs today.By buying an FPGA company, Intel could theoretically use knowledge it has in this area to outdistance itself from the remaining standalone FPGA vendor – by working to integrate multiple intelligent subsystems, each running its own embedded firmware and application software. Intel could integrate its Atom cores into the FPGA product line, get rid of the licensing fees around other cores (ARM or PPC), thereby reducing the solution cost for FPGAs containing processing logic.However, this may be a tough sell with customers who have ties and familiarity to their existing processor solutions (ARM, etc).
Also, if Intel owned a broad line of FPGAs, they could encourage vendors to use new CPU solutions (e.g. Atom for embedded / consumer) and could supply significantly reduced pricing on stand-alone FPGAs, in order to provide glue logic to compete with standard ASSP and ASIC vendors. Since system vendors strive for high degrees of IP and software reuse, Intel could really go after the strategy of being a one stop shop. This would be much more difficult to attain in embedded markets however, since OEMs don’t like to see a single company dominate the supply chain.
3. Process Technology: Altera’s use of 40nm at TSMC along with its roadmap to other process advancements would take effort to port to Intel fabs before planned synergies would be met. On the surface it seems that Altera is a better fit from a process technology perspective, since Intel has much more knowledge of TSMC’s fab and processes than it does of UMC and Samsung, because it has already ported Atom to TSMC. However Xilinx is rumored to be switching to TSMC for 28nm in the future.
4. Distribution: Intel currently gets the bulk of its microprocessor revenue from a small collection of PC OEMs. Expanding the line card may be challenging for Intel’s sales organization. Conversely, Intel could be interested in buyingbetter relationship with customers in adjacent spaces in their quest to push passed PCs to future revenue opportunities. Either way the customer overlap is not significant for Intel in the markets they are pushing hard on, namely mobile.
This is by no means an exhaustive list, and unlike the analyst in this article, I don’t believe Intel wants in on this market (they are after mobile sockets and the FPGA guys aren’t exactly inside smartphones and netbooks) but this was an interesting exercise in why the semi market is likely to remain largely fragmented over the longer term, with pockets that are resistant to consolidation.
Related articles by Zemanta
- Altera Announces Fourth Quarter Results Sales Up Strongly (eon.businesswire.com)
- Intel to Buy FPGA Vendor? (insidehpc.com)
- Intel Caught Cheating In 3DMark Benchmark (tech.slashdot.org)