Breaking News

Second Chance for Rick Tsai

View Comments: Newest First | Oldest First | Threaded View
User Rank
realjjj   5/8/2017 1:50:19 AM
In high end Mediatek needs GPU from a marketing perspective not LTE. The industry might think otherwise but to consumers LTE is not important, very few have access to very fast LTE, vast majority of reviews don't test it. Consumers need to see CPU, GPU, a decent ISP and that's about that. Mediatek has built an image for itself of having weak GPUs so that's where they need a win the most and marketing needs to focus. Android is also suffering vs iOS due to storage perf and by taking some risks in this area , one can create an advantage vs other SoC vendors- folks on Youtube test "CPU perf" by loading a mix of apps and games and timing it. Storage perf is quite important in such a test. In high end, area is less relevant. You either have the best or you don't sell. Saving 5mm2 while loosing 90% of the sales is not the right path. Maximize perf and thermals , forget about area.

The hardest part for Mediatek will be to deal with Qualcomm's illegal patent practices. They can keep ignoring it but that's not pain free.

It will also be interesting to see who their customers are. The device makers, the carriers or the consumer? There are conflicting interests between those categories, Qualcomm focuses on carriers while Mediatek has focused more on the other two so for- this is what hurt Mediatek the most  this year, China Mobile going Cat 7 while Mediatek's mid range solutions are only Cat 6.

For the future, AI related fields are the key to everything but they also need to be prepared for glasses as the new  form factor for everything computing and media delivery - phones, TVs ,PCs, consoles and so on get replaced by glasses. Qualcomm has a rather good GPU, they are involved with LETI on CoolCube so they seem better positioned for glasses for now. Qualcomm's weak point is the excessive focus on 4G and to some extent it is exploitable.

User Rank
10nm and 7nm pullback
resistion   5/8/2017 1:35:34 AM
MedaTek was one of those expected to drive 10nm and 7nm at TSMC, but the momentum seems to have gone out. Might be too risky for new executive situation.

Most Recent Comments
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed