Your browser is not supported. Please update it.

9 February 2017

Maxlinear juggles complexity effortlessly, adds Marvell’s G.hn

The surprise purchase overnight of the Marvell G.hn business for a paltry $21 million, by US RF chip upstart, Maxlinear has sent shockwaves through the home networking community. It also sends mixed messages about the future of multiple home networking technologies.

The first, rather skeptical thought we had was, “Is this a case of buying the competition to shut it down?” and that thought was waved aside by the fact that Maxlinear is only recently cash rich, and would never throw that money away when there are four or five other chip sources for G.hn. But the second thought was equally bad, that if the cost was only $21 million it puts an exceptionally low price on an asset which was built over a hard-fought 6 years or so, and Marvell can usually make money in markets which larger rivals have failed to make work. In the past, it has bought failed divisions from companies like Intel, and turned a profit on them, so why can’t it with G.hn?

The answer, we suspect, is all about appropriate sizing, although we won’t get to ask the companies about this until later this week. When we first came across Maxlinear it raised over $80 million in an IPO, less than the amount of cash that it has thrown off this year. Now it is larger and ready to take on more substantial adventures. It came to life by solving the ISDB-T 1 seg chip market in Japan, and went on to earn a living replacing TV can-tuners with CMOS chips. Now it splits its revenue and profits between the MoCA chips it inherited with the purchase of Entropic, and the tuners for advanced DOCSIS modems, which require full band capture, a trick only managed by Maxlinear and Broadcom to date.

In short, we think it wants G.hn because it sees a future in the powerline version, as HomePlug looks oblivion in the face over the coming year to 18 months. Maxlinear will retain MoCA because that technology already leads in the coaxial home network market. What kind of magic can it create with combination chips of the two technologies we can’t yet imagine, but one thing is for sure, Marvell will no longer pull together the G.hn community, and Maxlinear is unlikely to take over that role – and it may well lead to the rest of the community losing their way.

The fact that the deal includes the entire Spanish Marvell operation, gives the biggest clue, because it will have included the team of engineers who began life at DS2, a company which single handedly took on Homeplug with a superior powerline format, which was later sucked into the G.hn standard.

Maxinear gave details of the deal at its results, saying its Q4 numbers showed revenue of $87.1 million, a couple of $million above consensus, throwing off some $27.6 million of cash – less than it paid for the G.hn assets. For the year, its revenues were up 29% to $387.8 million, and free cashflow rose to $117 million.

Maxlinear now boasts some of the most complex analog and mixed-signal technology platforms across broadband – these are chips that other companies have trouble getting to work and even more trouble making work economically. To us, this is a company headed rapidly into the $1 billion revenue size – in little more than a year or two, and unless someone snaps it up, it will begin to challenge more broadly over that time, from its home networking base today. But this is not a company that wants scale, it is a company that wants to take on tough chip problems, and that drives margins high and makes it everybody’s friend.

However, this deal leaves a lasting question mark over why Marvell would give up the ghost on G.hn after such a long hard battle across the board against MoCA and HomePlug, just when it looked like it might be winning in at least the powerline end of market. Perhaps Marvell is looking to be single minded in offering G.fast and WiFi chips, and perhaps making more of an entry into cellular and wants to keep that focus. Perhaps? It could just as easily feel that G.hn will never be big enough to drive Marvell profits, but that doesn’t mean that Maxlinear can’t re-inject it with some enthusiasm and new ideas, and make plenty of money out of it.