The way I see the server cpu market in 2020, X86 will still command about half of all the chips sold, ARM will take about 25% at the low end, and IBM's PowerPC will take about 15% at the top-end with somewhere between 5 to 10 percent being manufactured by others.
The ARM server cpu market will be controlled 60% by the top two players. Number 2 will have a $1B business, and number 1 will have twice that. I assummed that there is no consolidation and that all 6 competitors are still around (there may eventually be some consolidation).
See the spreadsheet.
The ramp may not start until the Xeon competitive chips get out there. X-Gene 3 is sampling around H2'16, and shipping in H2'17.
Right now Chris, AMCC is poised to be number one or number two in the ARM server market in 2020. The reason I'm saying we could be number two is that CAVM is making huge gains in this area, and while it is troublesome, this still represents a 6-times increase in sales for AMCC, not including anything else. It's a great time to be a server cpu vendor with INTC on the ropes! One thing I would be on the lookout for is that AMCC in a fit of hubris does not get involved in a money-throwing contest with anyone with deep pockets. In other words, if it becomes time to sell, then SELL, take the money and walk away happy! I don't know enough about AMCC management to judge this.
Post by christopher on Jun 7, 2016 17:25:36 GMT -5
obviously, isn't now the opportune time for AMCC's marketing division to be operating at a very high level?
I mean this as the company getting its product signed up to be pushed out the door!
How does all of that work anyway? I have NO idea, word of mouth? advertising? if so, how?, where? or is it a sub-out type of situation?
with the #One ARM Server position pushing out around $2B in sales and the #Two ARM Server position pushing out close to $1B in sales. isn't this some serious product for a company the size of AMCC?
Would you be able to provide an estimate of share price for the #One and #Two players?
Last Edit: Jun 7, 2016 18:00:25 GMT -5 by christopher
I would assume the sales and marketing thing works the same way in most industries... make contacts at trade shows, contact customers directly, do demos, presentations, let customers do evaluations, etc. Then it all comes down to price and delivery.
The $1B or $2B number is ALL top-down. I would be very uncomfortable pegging share price to a particular market forecast. First, a good market forecast is within a multiple of 2X, so if the market is supposed to be $5B for ARM, then it could be anywhere from $2.5 to $10B -- who knows? There is also the aspect of timing and growth rate, which also figures into valuation quite a bit. If VTSS' products had gotten traction right away they might still be around, or at least lasted longer and been sold at a better price.
The bottom-up model I used previously argues for an overall market size for ARM server chips somewhere around $700M. Either somebody is wrong, or the ramp has not yet begun.
In good conscience, I cannot provide a definitive price projection at this time. It is my belief that the price will go up, I just don't know by how much.
I'm saying that AMCC will be around $880M in 2020 from ARM server CPU revenue, which is second behind CAVM (if they don't sell to someone else before then). For all the noise that Gopi has made about the eco-system, CAVM are the ones now pushing the eco-system forward through timely innovation and acquisition. While AMCC is moving the bleeding edge forward with PAM4, CAVM is moving the present ahead with one-stop shopping solutions for the ARM data center. CAVM will capture the biggest part of this market, which may leave AMCC with the high-end. Still, $880M is not too shabby. I adjusted the TAM down from $20B to $18B, which was initially adjusted from $15B up to $20B.