What does Marvell want to be?
Marvell has immense talent and a wide range of capabilities. But it is not clear how that all hangs together. It is hard to see Marvell as an independent company five years from now.
Marvell has immense talent and a wide range of capabilities. But it is not clear how that all hangs together. It is hard to see Marvell as an independent company five years from now.
There are a lot of reasons why Qualcomm is not a force in the custom ASIC business – fear of creating competing products, no sales culture – but we think the main reason is the can of legal worms it would open up for their IP licensing.
RISC V pioneer SiFive has gone through several iterations, but has now solidified a business model that essentially positions it as a direct competitor to Arm. If they can continue to execute they will benefit from much of the very healhty RISC V momentum
Apple’s antipathy towards Qualcomm pre-dates the launch of the iPhone and seems to be deeply ingrained in their culture. Qualcomm would probably be very flexible to normalize the relationship, but has limits to what i can offer, especially around IP.
Like the hero of a zombie movie – Intel has fixed its manufacturing process and awoken from its coma – only to find the world radically altered. Intel has a lot of talent, but the competition is clawing at the door.
Oppo is has largely given up designing its own chips, but they faced a very specific set of challenges which do not apply to many of the others looking to roll their own semis.
How do CPU companies market in a newly, highly-fragmented market? There are no easy answers in this very complex, rapidly moving environment, but there is some hope.
Apple has a history of relegating its homegrown chips to the Apple Watch. Is that the fate of their future cellular modem?
There are really three markets for AI semis – training, cloud inference and edge inference. All of them are already fairly crowded. Choose your battles carefully.
Many semis companies say they want to sell software. We think they really want recurring revenue and software company multiples. And what they really need is competitive advantage, which software can sometimes provide but is not the only way.