AMZN - Amazon.com Inc.

[Grenville]

If you wonder why the server market is tanking:

 

http://feedproxy.google.com/~r/OmMalik/~3/4qc8UjgeUTQ/

 

Thanks for posting the article. Interesting info from the conference and an insider at Amazon.

[Guest valueInv]

Article below is an WSJ interview with Andy Jazzy, head of AWS.

 

"Meet the Man Who Really Runs the Internet"

http://online.wsj.com/news/articles/SB10001424052702304868404579194353031011652

 

WSJ: What is the annual revenue of AWS? Is it profitable?

 

Mr. Jassy: You know I can't disclose that. We don't disclose our financial results. It will be a very significant free cash flow generating business for Amazon.

 

And you better believe that every single dollar of profit coming in from AWS is going right back into rapidly building out more capacity.  I saw one analyst saying that he thinks AWS has multiples of the capacity of the other large players when it comes to public cloud infrastructure.

 

However, the competition clearly is heating up.  I think that IaaS/PaaS providers who support both public, private, and hybrid cloud are going to be gaining market share.  People like Redhat are doing some pretty cool stuff -- OpenShift seems to be quite good.

 

But the pie is expanding at a rapid rate, so there's room for many players to win.

 

Which means the pie is shrinking for the server market. Note that he said that they pay 30% less

for servers. "Long runway" indeed.

 

No, there's really no way to definitively say that the pie is shrinking.  ASPs and margins are decreasing for sure.  But the units are expanding exponentially (just take a look at the non public cloud datacenter market growth). 

 

There's definitely a long runway.

The market is shrinking for the likes of Dell that don't make custom servers for the likes of Amazon. If Dell wants to enter this market, they will have to drop prices by more than 30% and reorganize their manufacturing to just start being competitive.

 

Look at the article I posted. The fact that Amazon is adding that much capacity in one day tells you how much of the market they are eating in a single day. That much computing has shifted from what would have been private computing to the public cloud.

 

This is not even counting Google, MSFT and others.

[Palantir]

Units may be expanding, but knowledgeable buyers like AmazFaceGoo can easily build their own servers. You know, the infamous "commoditization". So just tons of cheapo servers built by buying off the shelf equipment with no profit margin built in, so the pie may be increasing, but the profit pie is shrinking IMO.

 

The value here has to be the software that runs on the server, and thats where companies like AMZN, MS and RHT need to make their play.

 

[txlaw]

Units may be expanding, but knowledgeable buyers like AmazFaceGoo can easily build their own servers. You know, the infamous "commoditization". So just tons of cheapo servers built by buying off the shelf equipment with no profit margin built in, so the pie may be increasing, but the profit pie is shrinking IMO.

 

The value here has to be the software that runs on the server, and thats where companies like AMZN, MS and RHT need to make their play.

 

Yup, the direct from ODMs phenomenon continues to grow because of the big Internet cos.  It simply doesn't make sense for massive public cloud infrastructure/service providers like GOOG and AMZN to buy servers from the likes of DELL, HPQ, or IBM.  So commoditization is definitely occurring (and has been for a while) -- that's why you see ASPs and margins decreasing.  And revenues shrinking as well for what is considered the traditional server market.

 

However, it's important to realize that units are growing at a rapid clip with direct sales from ODMs not really encompassed within the unit count for most studies/reports (most server market share figures I've seen only concentrate on traditional server vendors).  The fact of the matter is that units are growing rapidly because of the amazing growth of remote computing.

 

What we cannot say is what happens to profits as the traditional hardware vendors -- the top three of which are IBM, HP, and DELL -- start providing their servers for a "converged cloud" world.  I happen to agree with IBM, HP, and DELL that we will be living in a hybrid world -- one where you have the option to go with public cloud, hybrid cloud, or private cloud.  And you can either have a managed or unmanaged cloud.  What type of solution you go with will depend on your business and organizational capabilities.

 

So think about this:  all of these guys are buying from ODMs.  DELL and HPQ are ODM buyers as well -- it's just that they make a retail profit on servers at this time.  They're actually in a better position than FB, AMZN, and GOOG in terms of buying power.  That means they can be low cost providers for non-public cloud solutions (think about verticals like healthcare or energy), and they can also provide their own public clouds.  Because of projects like OpenStack and OpenCompute.   

 

In order for these guys to totally lose out, you must assume that none of them will provide public cloud solutions on par with the current public cloud big dogs and that everyone will outsource to the public cloud -- or at a higher level, to SaaS providers.  I contend that because of projects like Open Compute and OpenStack, the public cloud playing field will be leveled over time, with resources, scale, and enterprise relationships starting to matter more and more in terms of competitiveness in the market.  I also think that there will always be enterprises who will go with managed or unmanaged private/hybrid cloud solutions.  Because these guys want control over their data, they want to outsource cloud infrastructure management to vendors with expertise, and because it actually costs less to buy than to rent in most cases.

 

In other words, "retail" server sales to third parties may shrink, but that doesn't mean that the big guys don't profit from their position in the long run, particularly when you see value shifting to the software and services bundled with the cloud hardware (again, I agree with the notion of commoditization).  That's exactly the approach that the big three are taking.

 

RHAT is ridiculously overvalued, btw.

[Palantir]

I'm a little unclear about the above. When you say IBM/HP/D will be building for the public cloud, do you mean to say their software platforms for the public cloud or becoming commodity suppliers for the hardware? If the latter, I don't see an advantage, but If the former, I suppose OpenStack is one way to do it, but its not obvious to me yet how it is going to be monetized. But I have to read more before I can have an opinion on this.

 

Focusing on the last part. Why do you think RHT is way overvalued? I think it's a steal.

[txlaw]

Focusing on the last part. Why do you think RHT is way overvalued? I think it's a steal.

 

To boil it down, it seems like the market is valuing it like a software company when it's really a services company.  And that's not good when there is so much competition coming online.

 

I would actually be interested to hear why you think it's a steal, as I haven't done a ton of work on the company. 

[Palantir]

My take - Their revenue has an annuity-like structure, their clients have high switching costs, they have a low margin business model that tends to be disruptive and hard for proprietary software to attack, and finally, their earnings are far less than FCF. Combining the solid business model with its high rates of growth, even if growth slows I still think it looks attractive (back out the cash and use FCF vs NI).

 

The issue though is what are they going to do after RHEL? There is only so much they can convert to RHEL, so they will need new lines of growth, and they seem to think they've found it with middleware. My opinion is that their business model tends to be pretty flexible and should be able to attack other spaces apart from just operating systems. But I'm not really a tech guy...

 

 

 

[Guest valueInv]

 

So think about this:  all of these guys are buying from ODMs.  DELL and HPQ are ODM buyers as well -- it's just that they make a retail profit on servers at this time.  They're actually in a better position than FB, AMZN, and GOOG in terms of buying power.  That means they can be low cost providers for non-public cloud solutions (think about verticals like healthcare or energy), and they can also provide their own public clouds.  Because of projects like OpenStack and OpenCompute.   

 

In order for these guys to totally lose out, you must assume that none of them will provide public cloud solutions on par with the current public cloud big dogs and that everyone will outsource to the public cloud -- or at a higher level, to SaaS providers.  I contend that because of projects like Open Compute and OpenStack, the public cloud playing field will be leveled over time, with resources, scale, and enterprise relationships starting to matter more and more in terms of competitiveness in the market.  I also think that there will always be enterprises who will go with managed or unmanaged private/hybrid cloud solutions.  Because these guys want control over their data, they want to outsource cloud infrastructure management to vendors with expertise, and because it actually costs less to buy than to rent in most cases.

 

In other words, "retail" server sales to third parties may shrink, but that doesn't mean that the big guys don't profit from their position in the long run, particularly when you see value shifting to the software and services bundled with the cloud hardware (again, I agree with the notion of commoditization).  That's exactly the approach that the big three are taking.

 

Let me see, AWS has cut prices 38 times in the last 7 years. Players like Dell and others will catch up and  join the game with Openstack, etc. That means that cloud services will get commoditized even more quickly. So let's see:

Hardware is getting commoditized

The OS is getting commoditized

Software is being replaced by services

Services is getting commoditized

 

I am beginning to sense a pattern here  ::).

Where is a poor boy to invest?

 

 

[txlaw]

I'm a little unclear about the above. When you say IBM/HP/D will be building for the public cloud, do you mean to say their software platforms for the public cloud or becoming commodity suppliers for the hardware? If the latter, I don't see an advantage, but If the former, I suppose OpenStack is one way to do it, but its not obvious to me yet how it is going to be monetized. But I have to read more before I can have an opinion on this.

 

I see you updated your last post.

 

What I'm saying is that, to the extent we're talking about "public cloud" service providers posing a competitive threat to traditional server vendors -- because many people who bought servers (e.g., Fortune 500 enterprise or SaaS providers) will now rent compute power (and other services) from AMZN, GOOG, or other public cloud providers who use direct from ODM hardware -- this market share loss is mitigated by: (1) exponential growth in cloud computing units needed by everybody (whether via private, public, or hybrid cloud); and (2) a shift in biz model where these big cloud hardware vendors start renting out the hardware they purchase and assemble from ODMs (for example, IBM and HP providing their own public cloud services) in addition to their traditional biz of selling hardware and providing turnkey solutions where hardware sales are part of a package.  Software and services provided to manage/utilize cloud computing infrastructure are part and parcel of both public and private cloud provisioning. 

 

Here's the way that NIST separates out the deployment models we're talking about:

 

(1) Private cloud -- The cloud infrastructure is provisioned for exclusive use by a single organization comprising multiple consumers (e.g., business units). It may be owned, managed, and operated by the organization, a third party, or some combination of them, and it may exist on or off premises.

 

(2) Community cloud -- The cloud infrastructure is provisioned for exclusive use by a specific community of consumers from organizations that have shared concerns (e.g., mission, security requirements, policy, and compliance considerations).  It may be owned, managed, and operated by one or more of the organizations in the community, a third party, or some combination of them, and it may exist on or off presmises.

 

(3) Public cloud -- The cloud infrastructure is provisioned for open use by the general public.  It may be owned, managed, and operated by a business, academic, or government organization, or some combination of them.  It exists on the premises of the cloud provider.

 

(4) Hybrid cloud -- The cloud infrastructure is a composition of two or more distinct cloud infrastructures (private, community, or public) that remain unique entities, but are bound together by standardized or proprietary technology that enables data and application portability (e.g., cloud bursting for load balancing between clouds).

 

Note that SaaS vendors can provide their software off of public cloud infrastructre (e.g., Netflix on AWS), private cloud infrastructure (e.g., Facebook), or hybrid cloud (e.g., Box.net).  The very big SaaS vendors like GOOG and FB tend to buy their own infrastructure straight from ODMs.

 

When it comes to private and community cloud (and hybrid cloud, as well), the software vendors that we're talking about are still making cloud hardware sales to organizations who need (or want) to own, rather than rent, the infrastructure.  But as I said before, they're also getting into the deployment models that focus more on renting compute hardware.  Further, it's not a zero sum game when it comes to the growth in public cloud computing, or the increasing use of commodity hardware by big SaaS vendors in their own private clouds.  The need for compute power is growing, and you are gonna see a lot of demand for compute power from all sorts of organizations that never considered this to be a necessity in the past.  The demand for compute power is on a hockey stick growth curve.

 

So the way I see it, there will be multiple winners.  AMZN will be one of them, of course.  And the big cloud hardware vendors will hopefully also be winners if they play their cards right.

[Guest valueInv]

I'm a little unclear about the above. When you say IBM/HP/D will be building for the public cloud, do you mean to say their software platforms for the public cloud or becoming commodity suppliers for the hardware? If the latter, I don't see an advantage, but If the former, I suppose OpenStack is one way to do it, but its not obvious to me yet how it is going to be monetized. But I have to read more before I can have an opinion on this.

 

I see you updated your last post.

 

What I'm saying is that, to the extent we're talking about "public cloud" service providers posing a competitive threat to traditional server vendors -- because many people who bought servers (e.g., Fortune 500 enterprise or SaaS providers) will now rent compute power (and other services) from AMZN, GOOG, or other public cloud providers who use direct from ODM hardware -- this market share loss is mitigated by: (1) exponential growth in cloud computing units needed by everybody (whether via private, public, or hybrid cloud); and (2) a shift in biz model where these big cloud hardware vendors start renting out the hardware they purchase and assemble from ODMs (for example, IBM and HP providing their own public cloud services) in addition to their traditional biz of selling hardware and providing turnkey solutions where hardware sales are part of a package.  Software and services provided to manage/utilize cloud computing infrastructure are part and parcel of both public and private cloud provisioning. 

 

Here's the way that NIST separates out the deployment models we're talking about:

 

(1) Private cloud -- The cloud infrastructure is provisioned for exclusive use by a single organization comprising multiple consumers (e.g., business units). It may be owned, managed, and operated by the organization, a third party, or some combination of them, and it may exist on or off premises.

 

(2) Community cloud -- The cloud infrastructure is provisioned for exclusive use by a specific community of consumers from organizations that have shared concerns (e.g., mission, security requirements, policy, and compliance considerations).  It may be owned, managed, and operated by one or more of the organizations in the community, a third party, or some combination of them, and it may exist on or off presmises.

 

(3) Public cloud -- The cloud infrastructure is provisioned for open use by the general public.  It may be owned, managed, and operated by a business, academic, or government organization, or some combination of them.  It exists on the premises of the cloud provider.

 

(4) Hybrid cloud -- The cloud infrastructure is a composition of two or more distinct cloud infrastructures (private, community, or public) that remain unique entities, but are bound together by standardized or proprietary technology that enables data and application portability (e.g., cloud bursting for load balancing between clouds).

 

Note that SaaS vendors can provide their software off of public cloud infrastructre (e.g., Netflix on AWS), private cloud infrastructure (e.g., Facebook), or hybrid cloud (e.g., Box.net).  The very big SaaS vendors like GOOG and FB tend to buy their own infrastructure straight from ODMs.

 

When it comes to private and community cloud (and hybrid cloud, as well), the software vendors that we're talking about are still making cloud hardware sales to organizations who need (or want) to own, rather than rent, the infrastructure.  But as I said before, they're also getting into the deployment models that focus more on renting compute hardware.  Further, it's not a zero sum game when it comes to the growth in public cloud computing, or the increasing use of commodity hardware by big SaaS vendors in their own private clouds.  The need for compute power is growing, and you are gonna see a lot of demand for compute power from all sorts of organizations that never considered this to be a necessity in the past.  The demand for compute power is on a hockey stick growth curve.

 

So the way I see it, there will be multiple winners.  AMZN will be one of them, of course.  And the big cloud hardware vendors will hopefully also be winners if they play their cards right.

 

Here ya go  ;):

 

http://gigaom.com/2013/11/16/new-startup-economics-why-amazon-web-services-and-dropbox-need-each-other/

 

Some interesting tidbits:

AWS has over five times the compute capacity in use than the aggregate total of the other 14 providers.

AWS replaces an average of 400 servers per customer.

Amazon has lowered AWS prices 38 times since launching in 2006.

 

In other words, thats 400 serves they don't buy from Dell, HP and IBM.

[Guest valueInv]

[

 

So think about this:  all of these guys are buying from ODMs.  DELL and HPQ are ODM buyers as well -- it's just that they make a retail profit on servers at this time.  They're actually in a better position than FB, AMZN, and GOOG in terms of buying power.  That means they can be low cost providers for non-public cloud solutions (think about verticals like healthcare or energy), and they can also provide their own public clouds.  Because of projects like OpenStack and OpenCompute.   

 

In order for these guys to totally lose out, you must assume that none of them will provide public cloud solutions on par with the current public cloud big dogs and that everyone will outsource to the public cloud -- or at a higher level, to SaaS providers.  I contend that because of projects like Open Compute and OpenStack, the public cloud playing field will be leveled over time, with resources, scale, and enterprise relationships starting to matter more and more in terms of competitiveness in the market.  I also think that there will always be enterprises who will go with managed or unmanaged private/hybrid cloud solutions.  Because these guys want control over their data, they want to outsource cloud infrastructure management to vendors with expertise, and because it actually costs less to buy than to rent in most cases.

 

In other words, "retail" server sales to third parties may shrink, but that doesn't mean that the big guys don't profit from their position in the long run, particularly when you see value shifting to the software and services bundled with the cloud hardware (again, I agree with the notion of commoditization).  That's exactly the approach that the big three are taking.

 

And here is Openstack in the real world:

http://blogs.gartner.com/alessandro-perilli/why-vendors-cant-sell-openstack-to-enterprises/

 

Re:Invent was a message to the ecosystem:

http://gigaom.com/2013/11/20/aws-gets-respect-but-also-fear-and-loathing-in-las-vegas/

[Guest valueInv]

[

 

So think about this:  all of these guys are buying from ODMs.  DELL and HPQ are ODM buyers as well -- it's just that they make a retail profit on servers at this time.  They're actually in a better position than FB, AMZN, and GOOG in terms of buying power.  That means they can be low cost providers for non-public cloud solutions (think about verticals like healthcare or energy), and they can also provide their own public clouds.  Because of projects like OpenStack and OpenCompute.   

 

In order for these guys to totally lose out, you must assume that none of them will provide public cloud solutions on par with the current public cloud big dogs and that everyone will outsource to the public cloud -- or at a higher level, to SaaS providers.  I contend that because of projects like Open Compute and OpenStack, the public cloud playing field will be leveled over time, with resources, scale, and enterprise relationships starting to matter more and more in terms of competitiveness in the market.  I also think that there will always be enterprises who will go with managed or unmanaged private/hybrid cloud solutions.  Because these guys want control over their data, they want to outsource cloud infrastructure management to vendors with expertise, and because it actually costs less to buy than to rent in most cases.

 

In other words, "retail" server sales to third parties may shrink, but that doesn't mean that the big guys don't profit from their position in the long run, particularly when you see value shifting to the software and services bundled with the cloud hardware (again, I agree with the notion of commoditization).  That's exactly the approach that the big three are taking.

 

And here is Openstack in the real world:

http://blogs.gartner.com/alessandro-perilli/why-vendors-cant-sell-openstack-to-enterprises/

 

Re:Invent was a message to the ecosystem:

http://gigaom.com/2013/11/20/aws-gets-respect-but-also-fear-and-loathing-in-las-vegas/

 

And here's more:

 

http://gigaom.com/2013/11/26/doh-amazon-builds-its-cloud-lead-over-it-rivals-in-q3/

[CONeal]

In a recent interview on 60 minutes Amazon revealed they were working on drone delivery for packages up to 5 pounds.  Thought it would be possible to implement in about 5 years.  Earliest that it could be done is 3 years due to Needing FAA approval. 

 

Edited to change FFA to FAA

[LC]

http://www.amazon.com/b?ref_=tsm_1_tw_s_amzn_mx3eqp&node=8037720011

 

If they can work it all out, it seems like a good way to deliver to rural customers. Although I suspect some money will be made insuring all the various liabilities associated with it...

[turar]

In Globe & Mail weekend edition there was an article about Chinese e-commerce. Couple of things mentioned were that one of Chinese e-tailers already tested drone delivery to rural areas (albeit with much less fanfare), and that by some estimates, China's online retailing is already larger than that of the US.

http://www.theglobeandmail.com/report-on-business/international-business/in-china-online-retail-is-about-to-get-a-whole-lot-bigger/article15688956/

 

[DCG]

China's population is 3x the U.S., so not a big surprise.

[Guest valueInv]

And the race to see who can lose the most money:

 

http://venturebeat.com/2013/12/03/google-cloud-compute-engine-launch/

[Guest valueInv]

In Globe & Mail weekend edition there was an article about Chinese e-commerce. Couple of things mentioned were that one of Chinese e-tailers already tested drone delivery to rural areas (albeit with much less fanfare), and that by some estimates, China's online retailing is already larger than that of the US.

http://www.theglobeandmail.com/report-on-business/international-business/in-china-online-retail-is-about-to-get-a-whole-lot-bigger/article15688956/

Alibaba is bigger than Amazon and Ebay put together.

[fareastwarriors]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

[Guest valueInv]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

 

They way they are doing it is problematic - using GPS co-ordinates. What are the coordinates of your front porch? How do you deliver to an apartment building? How valuable and stealable are the drones? How good are people at skeet shooting?

 

 

 

[Guest valueInv]

Something to think about:

 

http://www.businessweek.com/articles/2013-12-02/amazons-drone-fleet-delivers-what-bezos-wants-an-image-of-ingenuity

[rmitz]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

 

It's moving into different segments of the market.  You need some weird part for your plumbing project?  If it could be delivered in 30 minutes, why bother going to the store (if you can even get it there at all), just continue working on other bits while it's on the way to you.

[JAllen]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

 

It's moving into different segments of the market.  You need some weird part for your plumbing project?  If it could be delivered in 30 minutes, why bother going to the store (if you can even get it there at all), just continue working on other bits while it's on the way to you.

 

 

Also sales they lose to Walgreen's & CVS - urgent items: candy, cards, various urgent health items.  AMZN will be able to offer comparable delivery/acquisition times in thirty minutes.  Walgreen's is one of the few retailers I ever shop at aside from AMZN now, and it's only health and other time sensitive products.

[CONeal]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

 

You and me both.  Had a discussion with a friend the other day and he was literally about to flip out b/c his package wasn't going to be delivered Friday (UPS was closed) over a shirt.  People are losing the whole concept of patience these days or maybe its their expectations are way to high.  Wish I could figure out a way to accomodate these types of people b/c even though they are high maintence, I would charge them an arm and a leg to accomodate those demands (cause lets face it if your a pain in the ass I'm going to charge extra putting up with all the bs).

[Kraven]

I never understood why people need their stuff so quickly. The stuff I buy there is never urgent. If it comes in couple of days that's cool but if it comes next week that's okay too... Maybe I'm just weird...

 

I guess I am not their target for Prime (or Prime Air)...

 

You and me both.  Had a discussion with a friend the other day and he was literally about to flip out b/c his package wasn't going to be delivered Friday (UPS was closed) over a shirt.  People are losing the whole concept of patience these days or maybe its their expectations are way to high.  Wish I could figure out a way to accomodate these types of people b/c even though they are high maintence, I would charge them an arm and a leg to accomodate those demands (cause lets face it if your a pain in the ass I'm going to charge extra putting up with all the bs).

 

Could be a function of age and what you grew up with. Take smartphones for example or even just the internet. I can remember growing up and to find out the sports scores we would sit around and wait until the sports came on the 11 pm news. If for some reason you missed that or it was a late game you had to wait until you got the newspaper the next day. Sometimes when we really wanted to know we would call up the sports desk at the local paper and ask them but if we did it too often they got irritated. Now, if you want to know a score and for whatever reason the app on your phone doesn't connect immediately everyone freaks out.