The Intel Developer Forum wrapped up and there are a few weeks worth of presentations, keynotes and technical sessions to mull over. In the meantime, here are my top takeaways from IDF.
1. Intel is increasingly about software.
In recent months, Intel has acquired Wind River, advanced Moblin, a Linux-based operating system, and increasingly talked up software on its way toward 10 software related acquisitions.
At IDF (see special report), Intel highlighted Moblin in action:
Goldman Sachs James Covello noted Moblin’s importance:
The company demonstrated Moblin working with Ubuntu, the free Linux based operating system which aims to be a Windows alternative in several markets. While we believe that there are direct benefits to Intel’s efforts to optimize Moblin for x86, we believe the program is strategic as well. Historically, Intel has aggressively defended its position against potential competitors in its core microprocessor business. Recall that it entered the NOR flash, NAND, and GPU businesses when there was concern about current or future competitive MPU products from AMD, Samsung, and Nvidia, respectively. With Qualcomm now making a push with its Snapdragon product into the low-cost PC space and with the potential for Microsoft to port Windows to ARM-based platforms in the future, we believe Intel could potentially devote even more resources to Moblin in order to help deter Microsoft from supporting Windows on ARM.
Meanwhile, Renee James, vice president and general manager of Intel’s software and services group, gave some details on the chipmaker’s grand plans for Wind River. James said Wind River will operate as an independent subsidiary and Intel will support non-Intel gear. Wind River will run on all Intel-based platforms.
2. Intel is serious about solid-state drives in the data center.
Intel made it clear that SSDs are about to take off.
Covello reports:
Intel offered several presentations today on SSDs and several customer case studies to suggest strong ROI for SSDs in the datacenter. Intel believes that roughly 50% of workloads in the datacenter are currently performance constrained, and asserted that SSDs are compelling for these applications today, despite the price premium over HDDs. In 2009, Intel believes there will be roughly 40 million HDDs sold into performance datacenter applications where it thinks SSDs can offer significant benefits.
Also see: Intel’s Maloney: Enhanced technology can address demands of new data center
3. Intel thinks speed will matter for notebooks.
The quad-core processor market for laptops remains small, but Intel is poised to push the speed boundaries. Intel touted its Core i7 chips for notebooks as the fastest ever. Wells Fargo analyst David Wong say high-end notebook chips are important to preserve margins from the Atom netbook processors.
Also see: Intel brings Nehalem to notebooks
Wong also adds that it’s encouraging that Intel is hitting its roadmap targets:
Intel has updated its microprocessor list price to reflect the addition of three new 45nm quad-core notebook chips. These new chips, the Core i7- 920XM, 820QM and 720QM, are Intel’s first Nehalem notebook processors (Clarksfield), and the first products on the Calpella notebook platform. We think these chips help Intel extend its already considerable lead over AMD at the high end of the notebook space….AMD has not yet introduced any 45nm or quad-core notebook chips. The company’s fastest dual-core notebook processor runs at 2.4GHz with just 2MB of memory.
4. The server roadmap got a little love from Intel.
Intel talked up its server line-up, but CEO Paul Otellini didn’t focus on the enterprise too much. However, it’s good to know the enterprise roadmap.
Note that the company’s next generation Xeon will be socket compatible with the Xeon 5500 platform.
5. Intel has a solid portfolio of products in 2010.
IDF detailed Intel’s ability to cover multiple bases such as servers, a desktop refresh and thin-light notebooks as well as netbooks. Wedbush analyst Patrick Wang said:
We come away from Day 1 of IDF incrementally more positive on (1) a strengthening PC cycle – positive for revenue upside, (2) evidence that servers (Nehalem-EX), notebook (CULV), and desktop (Gulftown) are poised to see a richer product mix in 2010 – positive for margin expansion, and (3) crisp execution of product and manufacturing roadmaps.
6. Intel is still upbeat on WiMax—even as carriers eye LTE in the U.S.
Dadi Perlmutter, executive vice president of Intel’s mobility group, said:
Last year we introduced the first embedded WiMax that goes into notebooks. But having products with WiMax doesn’t matter. It’s all about having networks, because if you cannot connect, then it doesn’t matter. And we have networks being built in North America, in Russia, in Japan, and we have networks already there. We are building with our partners networks in other places like India, Malaysia, Taiwan. And many, many other places have all sorts of mobile and fixed WiMax all over the globe.
Perlmutter then called up Clearwire to demonstrate a video conference over WiMax relative to 3G. Cue the Clearwire commercials. Intel needs Clearwire to succeed or it will have a bunch of WiMax enabled chips with no place to go.
7. Intel expects to ramp up its graphics integration through 2011.
But analysts aren’t expecting a big impact on Nvidia just yet. Goldman Sachs analyst James Schneider concludes:
Intel provided few details regarding the progress of Larrabee, its discrete graphics product. While the company disclosed that development kits are currently shipping to a small number of strategic partners, there was, in fact, less technical detail offered on Larrabee than at past IDF meetings. Management provided a functional demonstration of Larrabee silicon, but at a performance level that does not suggest a product that is close to broad commercial release. This data, combined with our checks, leads us to believe that Intel is unlikely to emerge as a significant presence in the mainstream GPU market in the next 18-24 months.
8. Light Peak could be an interesting technology.
Wang says:
Intel unveiled an interesting technology called Light Peak with the hope of replacing all of the cables coming out of a PC with a single fiber optic cable – a venture with a similar end goal to wireless USB. While immaterial to our thesis on Intel, we note that the roll-out of Light Peak beginning in 2010 could impact the manufacturers of PC peripherals.
Given the rat’s nest of cables in my office I’m already a Light Peak fan. Intel outlines Light Peak in a blog post:
Intel will be working with the industry to determine the best way to make this new technology a standard and to accelerate its adoption on a plethora of devices including PCs, handheld devices, consumer electronic devices and more. The end goal is to make Light Peak a complement to existing I/O technologies by enabling them to run together on a single cable and at higher, and more scalable speeds.
9. Intel’s manufacturing prowess remains an advantage.
Otellini talked up the move to 22 nanometer manufacturing processes. The big message: Manufacturing still matters even as many of Intel’s rivals outsource it. You don’t get to 15nm process technology by relying on others to push Moore’s Law.
Also see: Intel: 32nm is just the beginning · Intel teases 22 nanometer chips · Otellini at IDF: Innovation shifts as technology grows beyond the PC
