Broadcom, Qualcomm, NXP – fish, chips and globalisation

Acquisitions are expensive in our industry: AOL paid $162b for Time Warner, Dell $67b for EMC, HP $33.6b for Compaq. The semiconductor market is especially full of M&A activity at the moment – just as Qualcomm is eating NXP, Broadcom is trying to eat it, although its latest $130b bid has been rejected. Qualcomm is paying $47b for NXP, making this the second largest ever acquisition in the IT industry and yet you may not know much about either, as they are chip designers making money from other IT manufacturers through OEM contracts. It’s an important win for globalisation and loss for the Netherlands and Europe. Regional and national regulators need to think about important parts of our market being owned by a single company or a few companies from a single country.

Qualcomm and NXP’s strategies have been similar – creating organic growth from increasing chip demand and buying smaller specialist design and software companies to address new areas and opportunities, although the former has been substantially more successful than the latter. See Figure for quarterly revenues since the beginning of 2003. Qualcomm has undoubtedly been playing in a bigger pond (mostly telecommunications) than NXP (mainly automotive).

Their origins were different, with NXP being born from Philips Semiconductor in 2006. Its largest acquisition was of Freescale Semiconductor for $11.8b in 2015. Qualcomm was an American startup, founded in 1985 by silicon valley academics and quickly developing a specialisation in CDMA processing for telecommunications. They have been different in their manufacturing strategies – Qualcomm has been ‘fabless’, while NXP has its own foundries.

Their employment strategies are also different: Qualcomm’s headcount growing organically and through acquisitions more or less inline with its revenue growth; NXP’s being more stochastic – its headcount declined from the total taken from Philips in 2006 and was boosted significantly by its Freescale purchase (see Figure).

Employees who joined Freescale in 2015 will end up working for their fourth company if/when Broadcom buys Qualcomm, if manage to avoid redundancy. M&A activity may cause less disruption to the semiconductor sector, since its basic business processes are simpler than many in and out side the ITC industry.

Net profit for the two companies have also been very different, with Qualcomm making more money for its shareholders in most quarters until recently (see Figure).

There is a similarity between semiconductor and mining industries: they both need to make huge capital investments – in fabrication plants and new mines respectively – and wait a number of years before they make any revenue (let alone profit) at all. In semiconductors there has been a on-going trend for suppliers to become ‘fabless’, designing new chips and licensing them to others for fabrication in a third company’s factories. Fabrication-only vendors include TSMC, UMC and GlobalFoundries.

Broadcom itself was built through acquisitions: Silverlake Partners bought Agilent (once HP’s test and measurement division), changed its name to Avago, which then bought Emulex and Broadcom before (confusingly for some!) changing its name from Avago to Broadcom (a reverse brand take-over).

The semiconductor market is of course a global one (the first chip costs you $3b, all the rest are free!) and its natural that the major players should seek advantages through buying their competitors. However regional and national regulators should consider the possible consequences of allowing unfettered consolidation. NXP is a Dutch company, so its acquisition – following UK ARM’s acquisition by Japanese Softbank – leaves STMicro (French) as the last major European chip company.

Having important specialised parts of the IT industry ‘owned’ by one or two vendors increases the possibility of component shortages when a single factory makes all chips of certain types for the world – and eventually of higher prices and slower technical development.