GaN power electronics will top $1 billion in revenue in a couple of years thanks to cross-fertilization with LED industry

LYON, FRANCE: Yole Développement announced its report “Power GaN – 2012 edition”. In this report, Yole Développement provides a complete analysis of the GaN device and substrate industry in the power electronics field – market forecasts, company involvements, etc.

Slow ramp-up but huge expectations
The GaN power device industry has probably generated less than $2.5M revenues in 2011, as only two companies (IRF and EPC Corp.) are selling products on the open market. However, the overall GaN activity has seen extra revenues as R&D contracts, qualification tests and sampling for qualified customers was extremely buoyant.Source: Yole Développement, France.

At very short term, IRF and EPC will likely remain the two main vendors of GaN power devices on the open market in early 2012. This market is likely to stay below $10M for devices, with the rest being made through R&D sales.

2013 should signal the transition from qualification to production ramp-up for several new entrants. The device market could reach the $50 million threshold. In 2014, most of these new entrants will ramp-up their capacity, and by 2015 the availability and adoption of qualified 600V+ GaN devices should see the market grow very quickly, and open doors to non-consumer applications. In 2015, 12-15 players will share the consumption of more than 100,000 x 6” (equiv.) epiwafers.

“Beyond that, if GaN is qualified in the EV/HEV sector, GaN device business could top the billion dollar line and the GaN-on-Si substrate market could exceed $300 million revenues by 2019,” explains Dr Philippe Roussel, business unit manager, Power Electronics at Yole Développement. However, it is still unclear how car makers will choose between SiC, GaN or the current silicon technology.

At the substrate end, R&D activities are still quite fragmented between several options involving GaN-on-Sapphire, GaN-on-SiC, GaN-on-GaN, GaN-on-AlN and GaN-on-Silicon. Nevertheless, GaN-on-Si is likely to take a dominant position as 6” is now available with more than 7μm thick GaN epi and 8” is under qualification. 8” diameter availability is probably the parameter that will make this technology choice obvious.

It is now obvious that the GaN power world attracts numerous newcomers. Yole Développement’s team has screened five companies positioned on the epiwafer business side and more than six GaN device pure-players, aside to another 15 Si-based power firms developing GaN technology.

GaN power electronics cross-fertilizes with LED industry
A new trend is LED players now starting looking at this new business opportunity and wondering how to put in place a strategy of diversification to convert their existing extra LED capacity into power. That represents an “epsilon” today, but Yole Développement assumes it may create some disturbances in the natural and organic expected growth

GaN power electronics past, present and future business is inseparable to the LED industry. Both are linked in technology and market dynamics:

* In the past, the premises of GaN epi technology came from the LED industry that has brought this technology from the labs to mass production.
* Today, the extensive developments of GaN-on-Si epiwafers fertilized both the LED and the power industry. Most of the epiwafer vendors are targeting these two segments with dedicated products and offers.
* Tomorrow, it is likely some incumbent LED pure-players will enter in the Power industry world, using their extra-capacity and existing tool-sets to make, at least epiwafers, or even power devices.

Thus, at the end of the day, Yole Développement will not talk about LED or Power sectors anymore, but rather about “GaN device industry” as main players could be the same.

A question of business model
Power device makers usually buy polished Silicon wafers, conduct the epi (or buy Si epi-wafers) if needed (FZ thin wafer doesn’t require epitaxy) then process the devices. This model is roughly the same for SiC technology.

For those who plan to enter in the GaN field, 2 scenarios could occur:
* Some may not integrate MOCVD GaN epitaxy. They will buy GaN epiwafers and process it in the existing CMOS Front-End lines, as they use to do with Silicon substrates (or SiC).
* Some will try to fully integrate the GaN process, from the bare silicon, the GaN epi and the Front-End.