Reports from multiple sources have indicated that Broadcom are downsizing their activities in the Small Cell market. While this doesn’t immediate affect products shipping today based on their design, we consider the implications and point to other recent developments.
Broadcom’s mainstream business
This US publicly listed company is a major player throughout important telecom markets, focussed on wireline, wireless and enterprise storage. It’s a huge and growing business with a market cap of $38 Billion, annual revenues of $3.8 Billion and has around $2 billion cash in the balance.
Their tagline is “Connecting Everthing” and you’ll find their chipsets embedded in DSL modems (at both ends of the wire), Wi-Fi/bluetooth/GPS, copper wire and fibre broadband transmission, storage disk controllers and smartphones.
They pick growth markets, invest both organically and acquire/licence technology, then reap the benefits.
Broadcom’s foray into small cells
Broadcom acquired Percello, a small cell broadband chipset startup, in 2010 for around $80 million. It was a good call at the time, when there were really only two femtocell chipset solutions - Picochip and Percello, and the latter won some major design wins including Ubiquisys and Sagemcom. We interviewed Greg Fischer, Broadcom’s SVP and GM for Broadband Carrier Access business at that time – and the future looked good.
Some companies are better than others at retaining and motivating staff who join through acquisitions. By all accounts, Broadcom did a good job of the integration and left the team intact to continue development. New chipset variants appeared, including products supporting both 3G and LTE (and software configurable between modes). Initially targeted at the very low cost residential market, where four or eight concurrent users was enough, we saw ever greater capacity introduced – more than enough for standalone Enterprise small cell products.
I positioned Broadcom’s market at the higher volume/lower capacity end of the spectrum, seeking to win and exploit in-building residential and enterprise markets rather than the outdoor very high capacity macrocell business. It was always going to be easier to gain design wins for new products and new startups using lower cost manufacturers such as set-top box/residential modem ODMs already familiar with their chipsets.
There have been some very significant wins – from Free France who are the largest and lowest cost residential femtocell deployment worldwide, to SpiderCloud who have both 3G and 4G for larger enterprise buildings.
Technically these chipsets operate well, the design is good and cost effective. Some competitors might say they are not quite as powerful (handling huge volumes of concurrent users), but I’d argue that they are fit for purpose for the market they are designed for. And my view is that we need to handle fewer concurrent users per cell, not more, so that everybody gets a larger slice of the pie.
I can only surmise that revenues from the addressable market don’t fit Broadcom’s investment profile and they’d rather put their money into other areas which promise higher returns. Network operators haven’t taken up in-building small cells (residential or enterprise) at the pace many originally expected.
No sudden stop
Users, operators, vendors and installers with products based on these Broadcom chipsets shouldn’t worry they will disappear overnight. Far from it. Other macrocell chipset vendors have downsized their R&D investments but continued selling and supporting their customers for many years. TI downsized considerably a couple of years ago – I recall I was the last analyst to be briefed by their small cell marketing team before it was disbanded – but continues to be a major component in Nokia’s successful Flexizone solutions.
Chipset investment lifecycles last many years, with heavy investment upfront that takes years before commercial products become reality and revenues start flowing in. Small cell vendors will also have made substantial investments in their designs which they’ll want to continue with. This high barrier limits newcomers quickly entering and disrupting the market.
Having made investments of over $100 million, Broadcom will almost certainly want to continue to manufacture their chips for as long as customers want to buy them (there’s little incremental cost to do so), and will likely retain at least a small support team to operate and ensure existing customers can continue as before. We may even see some firmware updates to fix or enhance existing products and tweaks to handle new 3GPP standards. But I doubt we’ll see any major new physical chip designs emerge, or any major PR announcement of this startegy change.
We saw a similar scenario with Picochip some years ago when they ran out of funds. Existing customers could continue to order chips and the support team answered queries, but the pace of new developments slowed. After being acquired by Mindspeed and later Intel, the situation improved. Intel invested in restoring good support service and reinstated a development program.
We’ve seen Intel dominate in TD-LTE small cell designs (an early decision by Picochip to open a Chinese R&D office which is still open today probably helped), and they seem to be winning back some market share again after almost being written off (e.g. Parallel Wireless selected them).
The only real competitor for Broadcom and Intel has been Qualcomm, which has patiently and slowly continued to invest and develop its own FSM chipset solutions. It likely their total investment in the small cell market would be measured in $100 millions to date. This has included investing directly in some of the leading small cell vendors - announcements have included Alcatel-Lucent (2013) and Airspan.
Qualcomm’s progress is beginning to materialise, such as at Reliance Jio in India, which has rolled out 10,000’s of Airspan small cells.
You’ll also see Cavium mentioned as a small cell chipset vendor. I’d position them at the very high end – they have extremely powerful products which are more akin to Intel’s large server chips, and more suitable for centralised datacentre servers than residential or even small enterprise designs. Think Cloud RAN connected by fibre everywhere rather than distributed standalone small cells.
The two major players in the macrocell basestation market have been TI (Texas Instruments) and Freescale. Both have downsized their investments in macrocell chipsets in recent years. After the big splurge on LTE macrocell rollouts, macrocell revenues are starting to decline this year and this will have an impact on the underlying chipsets. I’m sure that was something these companies were well aware of in advance, hence scaling back R&D investments in a timely manner.
Freescale was acquired by NXP and is currently under offer of $30 billion by Qualcomm. The basestation chipset business is probably a minor consideration in the deal, and I’m unclear how much (if any) synergy there would be between Qualcomm’s inhouse small cell chipset business and the older Freescale macrocell designs.
One the notable achievements of the Small Cell forum in the past has been to define a common interface between small cell chipsets and the software stack that sits on top. This FAPI (Femtocell API) specification makes it much easier to port software between chipset designs and avoids lock-in to a single vendor. While there are some differences and niggles, software stacks such as those from Radisys or Node-H easily support both.
This reduces the pressure on small cell vendors who may have products based on any one of the above silicon vendors. I know several already have designs in their labs using two or more alternative chipsets. Switching platforms may require a further approval test cycle with some of the larger network operators, but with so few chipset platforms this should be more of a formality.
Looking at the other side of the coin, we can see that almost all smartphones in the world are based on chipsets from four or five very large suppliers. Intel (who acquired Infineon a few years back) tried to enter that domain but recently decided to withdraw.
I’d note that we’ve also seen considerable consolidation of the larger RAN equipment vendors – there are now only three (Ericsson, Huawei and Nokia) with over 85% of the market.
This change in the small cell chipset market will in the longer term increase the share of the pie available to Intel and Qualcomm, improving the attractiveness of financial investment. Both are well positioned and have a lot to gain as and when the in-building small cell market grows.