"MACOM intends to divest the well-positioned but non-strategic Compute business within the first 100 days of closing."
I'm wondering who the computing business buyer is.
"MACOM intends to divest AppliedMicro’s well-positioned Compute business within 100 days from closing the transaction, as the business does not strategically align with MACOM’s long-term focus. AppliedMicro has been exploring strategic options for the Compute business and there is known strategic interest among several potential buyers and investors. MACOM will continue to support Compute customers and partners during this transition."
It may just be that there was no buyer who wanted both businesses. Macom is purchasing the connectivity business and arbitraging the compute business. The two businesses really go together (as CAVM found out), but finding a seller with this kind of vision might be difficult. I'm wondering if QCOM might be a buyer (of the compute business) after all.
I really am very interested to see who buys the compute business.
To me this validates that arm server processor market has not materialized. If you think it has, then where are the numbers? In addition the fact is they dont have the $ to develop the 10nm/7nm generation of chip. Another thing to note is the architecture change in the newer designs. The massive core count just indicates that they realize the "specialty server market" is dead and they are just trying to compete with Intel in the mainstream server market. I think its very likely the processor business gets sold for peanuts or closed down. macom would be dumb to continue spending 50% or operating costs on a product line only providing 1% of sales. They won't do it long and they understand they most likely get 0$ for processor. I'm sure they shopped around until they where exhausted and this was the best they could get.
Another option however is they already have a buyer lined up, but the price is so pathetically low; apm does not want to admit that defeat. Doing it this way may allow them to bury the processor buyout under the rug (it may be under material threshold for macom and acquirer hence terms can be kept private). This is just more speculation.
shares already unloaded. i'm not an arbitration player. i don't have any opinion on MTSI, but conceptually if you were buying apm to get exposure to the 'burgeoning' arm server market why would you keep that money in mtsi which is not in that area and has already committing to not getting into that area? I think its important in investing to avoid getting diverted away from your original investment thesis by circumstance; and in this case you could see how that can happen.
Another thing to keep in mind for a post-mortem on this investment. If the arm processor market was so big and apm so far ahead in development for this market, why no numbers? In this case i think it means both the market is not developing as expected AND they could not deliver. So how would you know this was happening? Looking at the numbers over the last 5+ years and ignoring the hype. The numbers can lie, but they are less likely to lie than a marketing person. i'm not being negative its not really a "lie" its hope and wishful thinking.
all things considered this sounds like the best buyout they could come up with. The development costs for those 10/7nm chips are so prohibitive that they really were cutting it close on being able to deliver the products before the money ran out. So they made a good move in taking this bid i think.
I think that a sale was one possibility, which I believe speaks to your point above, which APM, even though it is debt-free, does not have the deep pockets to go up against Chipzilla. I think X-Gene has the functionality, but it could take years to show the market that it also has the inter-oper-ability and the reliability to be on a par with Chipzilla's very long history with datacenters. The interconnect business was money, but the compute business is still a question mark, and while this (compute) is potentially the more lucrative of the two businesses, the years and deep pockets it would take is not a good match for a small company punching above its weight.
What I don't understand is why not sell the two businesses together? They are complementary, as CAVM demonstrated, and Gopi asserted. I think it must speak to the caution with which potential buyers must view the specter of competing directly with INTC: many have come before and failed. I assume there was no one buyer who was willing to take a package deal.
I would not be surprised to see a QCOM snatch up the compute business for next to nothing.
Oh well, I was "in" in the 5's and out in the 8's (60% gain... I would have to look at my entry and exit points after VTSS).
I might be interested to split between MACOM and whoever the compute business buyer is (it would have to be someone, like a QCOM with deep pockets). It could be a much larger player because the value proposition would be almost the same, since taking even a piece of INTC's monopoly hold on the datacenter market would still represent a sizable boost for even a large company. And ARM, I believe, still has the superior business model to INTC.
Let's see how this unfolds over the coming months.
Why invest in MACOM? A good company with good products, technology, and cash flow. I also think the consolidation in former telecom stocks is over. VTSS, PMCS, and AMCC are all sold. Any ideas?
Personally, I would find MACOM a good low risk investment, and whoever the likely much larger buyer of the compute business is an equally low risk (lower-ish return too) possibility.
The datacenter is the future (at least the immediate future).
"Commenting on the transaction, John Croteau, President and Chief Executive Officer, stated, "This transaction will accelerate and expand MACOM’s breakout opportunity in Enterprise and Cloud Data Centers. MACOM will now be able to provide all the requisite semiconductor content for optical networks - analog, photonic and PHY - from the switch to fiber for long haul, metro, access, backhaul and Data Center. AppliedMicro’s 100G to 400G single-lambda PAM4 platform should perfectly complement MACOM's leadership in analog and photonic components for Data Centers."
"MACOM intends to divest the well-positioned but non-strategic Compute business within the first 100 days of closing."
Interpretive statement:
MACOM does not see itself competing with INTC
There IS a buyer lined up (I wonder how much this will offset the cost of the interconnect business).
I think the why to this strange deal is that the valuation of the processor side is too low for apm board/pg to accept. This allows them to save face. Using an occam's razor approach seems the only likely solution. I really doubt Qualcomm buys the processor unit from macom. Look at eetimes.com article stating Qualcomm is in production with a 10nm SOC at Samsung. APM was not in production with their Xgene3 yet. APM is already behind, that's another reason for the deal now. The Q has all the resources it needs to deliver its own arm server for sure, they dont need anything apm has.
A more interesting question is who buys macom next year? The semis are in a massive consolidation right now because moore's law is over. I dont think macom is big enough to stand alone, but if they have enough value someone will buy them out. Guest
Post by christopher on Nov 22, 2016 14:23:08 GMT -5
phobos,
For reference: I recently checked out the price of Microsemi (MSCC) I believe when Vitesse was sold in March of 2015 to Microsemi, the share price of Microsemi was around $35 share. Today Microsemi is trading around $56 share. 1 1/2 year up over $20
Guys who cares about ARM in the data center. Its already a non-starter. Look at this article. The future is all about machine learning and AI. Understand what's happening. All the apps, cars, and everything will be pinging data centers with requests to identify something or learn something. It's all AI/ML related. I saw a demo of a redbot using tensorflow to identify objects like a fork; see that was done in a data center not on the bot itself. See the data center had to be used to process millions of images of forks so a network could "know" what a fork is; the data center took the jpeg picture and had to process it through the network to identify it as a fork. See the data center operators that "get it" understand that saving 5% on power by implementing ARM is a non-starter; the reason is that they are facing +20-50% increase in power to handle new applications that require massive data and processing to train neural networks and do prediction for infinite applications. I think this is all happening very fast. Think about facebook identifying you in pictures your friends post; they already are doing this type of thing and be assured its costing $ and power. So the future is in chips that will accelerate ML/AI applications. Look in that direction. Guest
I noticed that CAVM's share price has also languished. They have some specialty chips, but they too are shifting from power pc to arm within the same old paradigm. Could the future belong to GPUs (or something like them)?
Guest, about QCOM, the article you cite regards mobile. It does not reference the datacenter (please correct if I read this incorrectly). Also, if the future belongs to Nervana, then the discussion of an ARM processor for the datacenter might be moot for all players, including ARM and QCOM. MACOM might be a good place to garner value, since they now seem to provide everything but the processor. Finding a good alternative to Nervana might be an interesting investment idea, but that company would still have to face the bottomless pockets of chipzilla. MACOM might be a good place to be.
Yea QCOM is very quite about the server side, but there are articles out there describing that they have a server processor in the hands of early potential customers. see link below or search on Q website. It may not be at 10nm yet due to mixed signal design requirements for IO standards, but for sure its already at 16nm, which is Xgene3. Q is also a goliath. The whisper is they have been working on 7nm for more than a year; but of course its a much longer process than PG would have you believe. My guess is that it is a relatively small incremental development cost for Q to develop server side processor because of all the work they have done on mobile processor.
I think Nvidia and the like have already scored big in GPUs for ML/AI. It will expand for them but also there will be competition from the likes of Nervana like products.
At this point i'm thinking a good "high risk but potential high reward" strategy in semis is to buy up all the little guys. If you do good research reduce the buy targets to ones that have a reasonable "enterprise value" that the big guys could buy. The trick is picking the ones they would want to buy. But the consolidation will continue for sure so this strategy has that going for it. So coming full circle maybe macom is a good prospect for this strategy because what they have fits into a big guys needs; but i do not know.
The bummer about the Nervana article is that all the acquired companies they mentioned were pre-IPO; so no way to get in on the game.
The future may be in AI, but there will still be a need for processors, perhaps ARM, and otherwise. If a smallish company gets even a little piece of what INTC commands, it could be a good thing:
I'm even thinking that if MACOM can arbitrage the compute division for much more than they paid for it ($65M), then that alone might give MACOM shares a boost. Perusing smallish semi shares with good enterprise value nothing really jumps out at me. I may just hold onto MACOM.
Post by christopher on Nov 23, 2016 16:13:46 GMT -5
Well it looks like AMCC finished at the high of the day today.
That being $8.45 share, .09 above the supposed sale price of $8.36
phobos,
I am also thinking hard of holding onto MACOM. From what I have been reading, they too have a good chance of being sold within the next couple of years.
Happy Thanksgiving! Already cashed out yesterday. Better to just buy up MACOM shares directly (if you're waiting for the acq. to complete) than let these shares be tethered to $MTSI.
Lots of other opportunities to be had by parking your money elsewhere.