Listened to a call hosted by CSFB on ARM, below are some notes for those that care. $9984.JP, $SFTBY...

34% of processors in the world ARM based.
Since being acquired, significantly increased investment in R&D – 50% increase in engineering headcount.
Plan is for HC to continue to increase.
60% operating margin when acquired. Now 10% operating margin.

Before acquisition, were repurposing smartphone processors for other markets like TV, servers. Wasn’t a bad strategy because smart phone was really demanding market so processor was very scalable, low area envelope, great power consumption. Most markets weren’t perfect fit..

Now – building set of common building blocks – then building blocks optimized for each individual market. Networking team, server team, automotive team, mobile team, etc. So maximizing performance for each end market.

New family of AI accelerators just launched. Inference engines for smartphones, TVs, microcontrollers. For instance microcontrollers can use machine learning to minimize vibrations. Just started licensing last quarter.

Have been building a brand new business – IoT managed services. If you want to deploy large numbers of IoT devices that are secured on a network, need to secure it. Have a SAAS business to make this easier/more secure.

Provide secure connectivity, device management/software updates remotely…lots of data needs to be managed, some devices move around. Do data collection/aggregation but not analytics. $100 million revenue business, but think could be $2 billion a year market one day.

In smartphones, have seen an increase in number of cores. 8-10 cores increasing, causing blended royalty rate to increase. ASP of application processors has gone up significantly. Typical price used to be $20-$30 – now $50-$60 and see even more expensive. Royalty doubled.

See a lot of new technologies coming into smartphones. AI, virtual reality, pixel processing requires a lot of silicon. ASP should increase as embed more in chips. Also opportunity to increase royalty rate - 50 bps for each extension of new lines. 100 bps if embed AI product.

In almost every market, companies who are the number one OEM are looking to build their own chips. Software becomes increasing differentiator in each market – washing machines, aircraft, autos, etc. Trying to mimic royalty percentage vs. if they were buying a merchant chip.

In data centers, Graviton 2 chips being deployed in AWS – ASIC targeted at number of applications in AWS, great performance for some applications. Complete replacement of Intel here. Marvel chips being deployed in Azure – 50% of applications could be used in ARM based silicon.

In China, Tencent and Alibaba have started to deploy HiSilicon chips in their data centers (ARM based). Company also talked about Apple news in PCs, which was well covered by media...

Overall, my takeaway is the company has been investing in a business that takes 4-5 years for those investments to pay off and they are starting to see measurable success. I think ARM will increasingly do well in next 2-3 years given the timing of their investment cycle.