Monday, June 13, 2011

Altair Semiconductor - Last (base)Band Standing?

Two years ago, when WiMAX still had a significant lead and LTE was but a distant 4G vision, Stanford professor and Beceem co-founder Dr. Arogyaswami Paulraj shared his thoughts on the wireless baseband processor business:
Dr. Paulraj expressed a concern for the scale advantage that the major semiconductor companies have in LTE, with Qualcomm and ST-Microelectronics accounting for more than 20,00 people, whereas the top three WiMAX vendors altogether have only ~500. (IEEE Mobile WiMAX Symposium, July 9/10 2009 - Napa, CA) 

Dr. Paulraj has been proven to be quite prescient, as today Beceem is no more, having been acquired by Broadcom. Shortly after that event, Wavesat filed for bankruptcy, and was subsequently acquired by Cavium Networks. Sequans struggled to pull off an IPO, but has seen its stock rise in the wake of the recent M&A activity, which includes Nvidia's acquisition of Icera which was completed today.

One of the few remaining independent baseband processor companies is Altair Semiconductor - based in Israel. Eran Eshed, 4G and LTE expert, Co-Founder and VP Marketing and Business Development at Altair Semiconductor recently spoke with EE Daily News to share his insight on developments in 4G and the baseband processor business.

The 4G market

At a macro-level market perspective, the single biggest event has been Verizon's rollout of LTE, and the traction it has established. Verizon shipped approximately 500,000 LTE devices in Q1, with about half being the HTC smartphone sales over just two weeks. We are seeing a lot of increased activity in emerging markets such as India and China. In Japan, Softbank launched their LTE trial recently.

Altair's business
At Altair, we have ~30+ design-ins now, with ~16 customers. Products range from USB dongles to larger CPE (customer premises equipment) routers, and even one smartphone (samples now, currently in test, targeting Q4 to market). We have shipped about 100,000 LTE chipsets to date, which makes us #2 behind Qualcomm. One of the unique things about our product offering is that it covers FDD (Frequency Division Duplex) as well as TDD (Time Division Duplex), in any 3GPP band worldwide. We have one product SKU, and with very minimal changes you can go to market in different markets with different carriers in different bands, minimizing R&D efforts. This is important for ROI because we are in the early days of the LTE market. The Taiwanese ODMs typically look for 200-250K units to start with before they kick off a new development.  Our products allow them to aggregate volume based on a single investment.

Competitive Landscape

In the FDD-LTE market we see Qualcomm, Samsung,and LG... that's about it. Broadcom, Mediatek, Marvell, Infineon, are far behind. On the TDD-LTE front, we see ourselves, Qualcomm, St-Ericsson and Huawei subsidiary HiSilicon.There aren't too many players today that have solutions that can actually ship. We've figured out the supply side, and now we are waiting for the demand side to pick up.

Thoughts on an IPO or M&A

Our competitors had a reason to liquidate. Sequans had probably maximized their WiMAX position. Beceem's investors were not willing to put any more money into the company. Icera also ran out of money, they were losing about $40M/year in cash flow. None of these companies, with valuations between $300-$400M, had a real LTE product. They had a 3G position and an LTE "potential" that got them these valuations. Our position is that LTE is just starting and this market is set to explode.This is not a time to liquidate, neither IPO or M&A. The valuation is growing. We just recently completed a Series-D of funding for $26M, so we're not in a rush to make any strategic moves. We're starting to build a good revenue pipeline.

We started with a WiMAX play, but we saw that it was not going to be a good opportunity for us, so we switched. We've been investing in LTE since 2006. Sometimes it's frustrating that we invested in WiMAX with no return, but if we hadn't gone through our WiMAX and XGP experience (for a Japanese carrier), we wouldn't be in the position to have a LTE product ready when the networks were being deployed.

Why did other WiMAX companies have difficulty transitioning to LTE?
With each new design generation, you are looking at a $30M investment best-case if you are efficient and experienced. The other companies just ran out of steam. Because of our SDR (Software-Defined Radio) approach, we had to spin less chips and we had shorter time-to-market. We just had to figure out the performance envelope and boundary conditions.

For our LTE product, it starts with the same architecture we used for WiMAX and XGP. Maybe 10-15% of the LTE product is a re-use of WiMAX. It's a new product and it takes less time because of our experience. LTE is more complex, and has much higher throughput and power consumption is an issue. LTE has about 15 different bands that are deployed across the world today, whereas WiMAX has just two. It's a different ballgame. It's tough.. semiconductors are expensive and it's tough to survive in this environment.

Challenges of LTE vs. WiMAX in a handset
LTE is not inherently more power consuming than WiMAX, both are less efficient than CDMA. The first difference is that you need speedier clocks to support the higher throughput. The second difference between LTE and WiMAX is that most LTE markets are FDD - you transmit and receive at the same time. Most WiMAX markets are TDD. Most competitors just license off-the-shelf DSPs. Competitors like Qualcomm use a hardwired DSP, so they are not as flexible. We provide a good combination of programmability and flexibility, together with low power consumption. That is much more critical in LTE.

Integration of Baseband and Application Processors
Qualcomm made nice inroads with their Snapdragon processors. Nvidia may also be going after that market segment. With the scale of this market, you can justify a diverse portfolio when you ship tens of millions of units. I don't see a (3G/4G) multimode integration happening so fast. Qualcomm is out there with an integrated device, but what we can offer is a low-cost high-performance dedicated LTE device that can sit side-by-side with a commoditized 3G chip. If you compare the combination of an Altair LTE chip with a Qualcomm MDM6600, you will find that it's a lower cost solution than an MDM9200 or MDM9600 from Qualcomm. There are a lot of advantages in looking at separate subsystems, like combining LTE with a lower cost Samsung or TI OMAP application processor, rather than integrating all in one. Then there is the single-mode LTE market. The single-mode market generated $250M for WiMAX chip providers last year. If you have a multi-mode device like a Qualcomm or an ST-Ericsson you are at a disadvantage to carry the "3G hump" on your back.

(interview conducted and edited by Mike Demler on June 9th, 2011)

No comments: