Broadcom is planning to phase out its Wi-Fi chip business?

[Marc66]

Communications chip maker Broadcom is planning to phase out its Wi-Fi chip business in order to streamline its workforce and product line, according to industry sources in Taiwan: http://www.macrumors.com/2016/03/18/apple-supplier-broadcom-wi-fi-chip-business/

 

[RMerlin]

I don't see any fact to back up this rumour, and I'm a bit skeptical. The wifi chip market is one that's expanding, as everything is becoming connected. Abandoning this market right now would be a mistake (and also bad for the market, who would see a major competitor disappear).

 

[sfx2000]

It's a Digitimes report - gossip and rumors from Taiwan and China

Even a blind squirrel gets a nut once in a while

 

[avtella]

I can't imagine Avago purchasing them (Broadcom) only to phase out the WiFi chips, maybe they'll be consolidated into the parent company?

 

[stevech]

I recall years back when Intel also announced it will exit from WiFi chips. Too little gross margin in it.

 

[System Error Message]

broadcom makes a lot of profit from wifi chips. This is because they make SoCs so when you use their SoC you would usually pair up a recommended broadcom wifi chip.

If broadcom gets purchased this could be possible but broadcom's wifi chip are usually quite good so it would be a big loss to the consumers. I also remember broadcom talking about innovation so their wifi chips would be used almost everywhere.

 

[mstombs]

If broadcom gets purchased

Avago Technologies acquired Broadcom in 2015, they still trade as Broadcom.

http://www.smallnetbuilder.com/wireless/wireless-news/32726-broadcom-gets-bought

Perhaps they are admitting defeat to Qualcomm Atheros in chasing next gen wifi, as demonstrated by Asus mixing Broadcom and Quantenna in recent top end routers

 

[System Error Message]

quantenna isnt qualcomm, a lot of people tend to make that mistake.

I always liked broadcom wireless chips because they provided throughput stability while atheros provided range.

Quantenna being a smaller company rushed their wifi so they got MU-MIMO routers first, The problems with their designs was that it was bottlenecked so it wouldnt work great if you wanted the maximum performance. Some of the issues users with the AC87U had was that overclocking didnt help throughput as much and didnt help improve wifi speeds or wifi association speeds.

Using Quantenna chips only for wifi is fine but as long as you dont follow their designs for the rest of the system or their SoC assuming their accelerators are in the chip.

Qualcomm's solution is their IPQ SoC with Atheros wifi chips which does better than broadcom's poultry dual core A9.

 

[RMerlin]

The problems with their designs was that it was bottlenecked so it wouldnt work great if you wanted the maximum performance. Some of the issues users with the AC87U had was that overclocking didnt help throughput as much and didnt help improve wifi speeds or wifi association speeds.

No, the actual problem is their interoperability with clients is pretty bad. Lots of compatibility issues with all kind of devices: iOS devices failing to connect, Android devices experiencing a major battery drain, etc.. And after two years, they STILL haven't produced a single stable and reliable driver.

Nobody complains about the performance itself - when it works, the speed is good. But the problem is it doesn't work for a lot of people, and their MU-MIMO support is over 12 months behind schedule.

 

[sfx2000]

Quantenna being a smaller company rushed their wifi so they got MU-MIMO routers first, The problems with their designs was that it was bottlenecked so it wouldnt work great if you wanted the maximum performance. Some of the issues users with the AC87U had was that overclocking didnt help throughput as much and didnt help improve wifi speeds or wifi association speeds.

Using Quantenna chips only for wifi is fine but as long as you dont follow their designs for the rest of the system or their SoC assuming their accelerators are in the chip.

Lots of folks bang on Quantenna (myself included) based on the consumer experience with Asus, Linksys, etc...

I've got a DirecTV Wireless Video bridge in the house as part of a recent installation (Genie DVR, two Mini Genie Wireless STB's, and the WVB), and guess what - the WiFi chip inside the WVB is -- Quantenna!

It's their older 4*4:4 802.11n (N600 class for 5GHz only), and it's stable and fast - but it's also in a very specific application doing a specific purpose - MOCA on one side, and WiFi on the other - there's nothing to keep me from probably attaching other clients besides the wireless STB's I suppose... it's strong as heck, as I'm getting 300+ Mbit/Sec throughput from the WVB to the minibox in the office, which is diagonal from the other corner where the AP is - My WRT1900ac can't reach that far in 5GHz... and it parked itself in DFS land automatically after the initial power up and OBSS scan (CH108 40MHz -1) - cool stuff...

I'll probably do a quick writeup on the DirecTV setup/installation from an SNB perspective if folks are interested...

 

[thiggins]

Quantenna's bread and butter has been application-specific, usually media streaming for service provider STB. When they control both ends of the link, performance is impressive.

But, as SFX notes, interoperability with the great unwashed msses of STAs has been its bete noire...

 

[System Error Message]

Well you may need to put notes that quantenna isnt related to qualcomm. Qualcomm's network solution is atheros whereas their recent SoCs for networking are IPQ processors (used to be atheros).

Broadcom has lots of investments in innovation in various areas and wifi is very important so they shouldnt be dropping out. A lot of the tiny stuff they make such as a mentioned coin cell battery with IC to perform tasks would also need wifi too. Even though they may not be getting as much profit from wifi, wifi tends to become an integral part of many other things.

 

[thiggins]

Getting back to the thread topic. This makes no sense. When Broadcom exited baseband cellular in 2014, they had no market share and were trying to break into a very competitive business. In the end, they were losing too much money and decided to cut their losses.

Broadcom has #1 or #2 market share position in Wi-Fi, depending on the segment. Margins would have to be tanking pretty badly to walk away from that sort of business.

 

[thiggins]

Perhaps they are admitting defeat to Qualcomm Atheros in chasing next gen wifi, as demonstrated by Asus mixing Broadcom and Quantenna in recent top end routers

What routers are you referring to? The ASUS RT-AC5300, RT-AC88U and RT-AC3100 are all all-Broadcom designs.

Quantenna hasn't been designed into any routers since the original AC2350/2400 designs.

 

[sfx2000]

I'd take the digitimes report with a grain of salt - WiFi is easy money for Broadcom/Avago - I can't see them walking away from the market position and investment they've made over the years...

 

[sfx2000]

Well you may need to put notes that quantenna isnt related to qualcomm. Qualcomm's network solution is atheros whereas their recent SoCs for networking are IPQ processors (used to be atheros).

Qualcomm had Airgo, Ubicom, and their own internal efforts already in place when they approached and did the M&A with Atheros...

Atheros had a great baseband, but running MIPS, they were running out of headroom - Qualcomm had a decent baseband, and a great core (Snapdragon ARMv7) - it really was a Reeese's moment - chocolate and peanut butter...

 

[mstombs]

What routers are you referring to? The ASUS RT-AC5300, RT-AC88U and RT-AC3100 are all all-Broadcom designs.

Quantenna hasn't been designed into any routers since the original AC2350/2400 designs.

Sorry the issues with the Quantenna firmware in the Asus AC-87U seem to generate too much noise for only one AC2400 model - which appears to have been a mistake and best avoided!

Broadcom also make the SOC for the RaspberryPi, I'm sure they originally were supplying an effectively obsolete set top box chip to a charitable cause at cost, but must be more serious now selling millions of quad-ARM-core 1.2GHz chips. The latest has mature Broadcom wifi as well. No use for a router though not Gigabit, and there's a USB2 bottleneck on connections.

 

[System Error Message]

Qualcomm had Airgo, Ubicom, and their own internal efforts already in place when they approached and did the M&A with Atheros...

Atheros had a great baseband, but running MIPS, they were running out of headroom - Qualcomm had a decent baseband, and a great core (Snapdragon ARMv7) - it really was a Reeese's moment - chocolate and peanut butter...

What i mean is network interface solution (such as for ethernet NICs, wifi cards/chips, etc).

qualcomm kraits (which are in snapdragons) are usually better versions of the equivalent ARM that is used in that time. A snapdragon would have more than just the IPQ so the IPQ isnt a snapdragon. Some snapdragons use stock ARM designs so unless the chip has the more complicated ARM, the kraits are a better choice. Some low end phones with snapdragon have ARM A7s while some use kraits.

The network interface market is rich because it is used everywhere regardless of medium. Broadcom usually make good NICs and they seem to be doing so well so i dont think they are going to leave the wifi market. Usually if this were to happen you would find many articles mentioning it not just one. Its more like the particular media wanted more hits.