You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Quarterly Highlights
• Revenues of $824 million, up 14% year-over-year, with record royalty revenue and strong growth in licensing revenue.
• Royalty revenue of $470 million, up 11% year-over-year was driven by the semiconductor industry recovery and the rapidly increasing penetration of Armv9-based chips, which typically command a higher royalty rate.
• Better than expected license revenue of $354 million, up 18% year-over-year, and strong bookings were due to strong demand for more advanced Arm CPUs as companies increase investment in AI across all end markets.
• Non-GAAP operating profit increased 17% year-over-year to $338 million resulting in a 41.0% non-GAAP operating margin.
It could be Google and Microsoft for cloud server CPUs. $140M is not a lot of money to those two. Arm cores are also in the IPU Intel sells to Google, and volume could be ramping up.
Wasn't there a story about ARM's pre-IPO plans to start charging royalties 'á la Qualcomm' as a percentage on the final end equipment price rather than the actual chip containing the ARM IP price ? I remember hearing quite a lot about this, together with much speculation about how much licencee resistance this would generate.
Either way, I'm curious if this a factor. Or if there's some other one off accounting/revenue recognition effect here. It's certainly very strange that the IPO was made at about $64 and the current stock price is around $121 - almost double after today's 57% (at time of writing) bump. That was only 6 months ago.
Wasn't there a story about ARM's pre-IPO plans to start charging royalties 'á la Qualcomm' as a percentage on the final end equipment price rather than the actual chip containing the ARM IP price ? I remember hearing quite a lot about this, together with much speculation about how much licencee resistance this would generate.
Either way, I'm curious if this a factor. Or if there's some other one off accounting/revenue recognition effect here. It's certainly very strange that the IPO was made at about $64 and the current stock price is around $121 - almost double after today's 57% (at time of writing) bump. That was only 6 months ago.
How would that pricing plan look for cloud servers? They don't have a price, only a cost.
My curiosity has always been about Architecture licenses. How much they cost, do they have a unit volume factor, etc. Ampere has announced, for example, they are moving from Arm IP cores to designing their own cores.
I still find it difficult to get excited about the Arm business model. RISC-V is going to eat much of the embedded business, and cloud server CPUs are not really high volume chips, by mobile, handheld, and desktop standards.
A friend of mine and I were discussing this Arm forecast and stock price bump (it's down about 7% from yesterday's close as I type this), and he thinks this enthusiasm is related to Nvidia's attempted acquisition of Arm a while back. His view is that a combination of Nvidia's attempt and Arm stating that AI chip companies are interested in advanced Arm CPUs (which I assume means v9), means that Arm is now an AI bubble company, and hence the ridiculous valuation. I think this theory makes some sense.
They have little choice. Their prospective customers can license a full 64-core design direct from ARM, slap on some IO and memory chiplets, and presto! a server chip fit for a cloud. Where is Ampere's leverage without a superior core?
They have little choice. Their prospective customers can license a full 64-core design direct from ARM, slap on some IO and memory chiplets, and presto! a server chip fit for a cloud. Where is Ampere's leverage without a superior core?
I completely agree. I suspect Amazon's Graviton is derived from the Arm design you mention. But as we both know, designing a competitive core for the Arm instruction set is not easy. Since I know a lot of people who work there, I hope Ampere has what it takes.
The only thing stronger than having absolute control, as happens in monopolies and oligopolies, is the strength that comes from numbers. Both are economic
A friend of mine and I were discussing this Arm forecast and stock price bump (it's down about 7% from yesterday's close as I type this), and he thinks this enthusiasm is related to Nvidia's attempted acquisition of Arm a while back. His view is that a combination of Nvidia's attempt and Arm stating that AI chip companies are interested in advanced Arm CPUs (which I assume means v9), means that Arm is now an AI bubble company, and hence the ridiculous valuation. I think this theory makes some sense.
It's also worth remembering just how small the free float is for ARM after the IPO - SoftBank still owns almost all the shares and most of the rest are likely held by the corporate investors (ARM's big licencees). It's a perfect scenario for share price volatility (up 55% late last week, down 18% yesterday). Almost as if that were the plan ...
I really do wonder just why investors (and regulators) put up with this sort of thing from tech companies. Alongside allowing founders like Mark Zuckerberg to hold onto majority control with a minority shareholding.