Understanding Changes in the Software & Venture Capital Industries

Posted on Jun 28, 2011 | 36 comments


The Venture Capital industry has changed over the past 5 years that I would argue are a direct result of changes in the software industry, not the other way around. Specifically, Amazon has changed our entire industry in profound ways often not attributed strongly enough to them.

I believe the changes to the industry will be lasting rather than temporal change. Venture capital is in the process of its own creative destruction with new market entrants and new models of innovation at the precise moment that our industry itself is contracting.

I will argue that when the dust settles, although we will have fewer firms, each type well end up more focused on traditional stage segments that cater to the core competencies of that firm. The trend of funding anything from the first $25k to funding $50 million at a billion+ valuation is unlikely to last as the skills and style to be effective at all stages are diverse enough to warrant focus.

I will argue that LPs who invest in VC funds will also need to adjust a bit as well.

Rewind
When I built my first company starting in 1999 it cost $2.5 million in infrastructure just to get started and another $2.5 million in team costs to code, launch, manage, market & sell our software. So it’s unsurprising that typical “A rounds” of venture capital were $5-10 million. We had to buy Oracle database licenses, UNIX servers, a Sun Solaris operating system, web servers, load balancers, EMC storage, disk mirrors for redundancy and had to commit to a year-long hosting agreement at places such as Exodus.

Open-Source Software & Horizontal Computing
The first major change in our industry was imperceptible to us as an industry. It was driven by the introduction of open-source software, most notably what was called the LAMP stack. Linux (instead of UNIX), Apache (web server software), MySQL (instead of Oracle) and PHP. Of course there were variants – we preferred PostGres to MySQL and many people used other programming languages than PHP.

Open source became a movement – a mentality. Suddenly infrastructure software was nearly free. We paid 10% of the normal costs for the software and that money was for software support. A 90% disruption in cost spawns innovation – believe me.

We also benefitted economically from a move to “horizontal computing.” What this meant was that rather than buying really expensive UNIX servers (and multiple machines in order to handle redundancy) we could buy cheap, replaceable servers for compute resources.

As our needs grew we could just add more cheap boxes and as boxes failed we could just chuck them out. We had to learn how to be better at “load balancing & replication” – meaning how we managed data across all the boxes since they weren’t centralized on one box.

These two trends had a major impact on the computing industry from 2000-2005 but the effects weren’t yet felt by the VC industry.

The Emergence of “Open Cloud” Infrastructure
The biggest change in the software industry beyond open-source was “open cloud.”

When we talk about cloud computing we have to be careful to differentiate between open cloud (services the are provided solely to for the economic purpose of building a cloud business) and the “platform cloud” where certain service providers offer cloud services wrapped around their core product. These are very different.

Platform cloud players like Salesforce.com provide compute resources so that third parties can build applications that integrate with its core product. That’s awesome for users of Salesforce.com or companies that want to cater to them but less awesome for pure startups that want independence and are really just looking for cloud infrastructure. Facebook is a “platform cloud” provider, too. That makes both of these amazing companies great channels for startups.

True that Salesforce.com in particular has made interesting moves toward open-cloud services by purchasing Heroku and also launching Database.com. It seems if anybody wants to move more toward open it will be Salesforce.

But for now when you want to build an independent, high-growth, VC-backed startup you need to build your overall company on a truly open cloud.

Enter Amazon.

They came from a different perspective. They have the mass retailer mentality of “stack ‘em high and sell ‘em cheap.” They started by offering cloud storage (S3) on a super cheap, pay-as-you consume basis. Every startup I knew in 2005 (when I started my second company) was using this. Why would we commit hundreds of thousands to EMC before we knew whether we had a big business?

They then launched processing capabilities (EC2) and we startups suddenly didn’t need to buy production servers. Then they launched a simple database, management tools and so on. Amazon will surely keep moving up the stack. My bet is that they fold A9 (their search tool) into AWS and offer search-as-a-service, too.

It sure would put pressure on Google if they had Facebook competing on one side of them for share of users’ time and Amazon flanking them on the other side by providing search to every website out there that might threaten AdSense and even Google’s core search business. Who knows?

If you want a deeper understanding of the layers of the cloud , how it is emerging and some of the exciting new players you can read it here.

Amazon changed our industry. This is mind boggling. That little online book company. Not Google. Not Microsoft. Not IBM, HP, Accenture, Cisco, Salesforce.com or anybody else. Amazon. 100% of the credit. And 9 years after they launched AWS there are still no credible competitors.

I find this strange. And maddening.

That said, Amazon – through AWS – even without strong competition is as wonderful an experience as Amazon the eCommerce retailer feels to you as an online shopper. Jeff Bezos simply deserves to be held up with Steve Jobs as two of the most important people driving innovation in computing today.

Spawning of Micro VCs
The biggest media attention in our industry went to the so-called “super angels” during the 2009/10 timeframe and while I don’t believe there is such thing as a super angel I believe that much media attention was deserved.

The earliest people that I spoke to who understood the changes in our industry were True Ventures & First Round Capital. They built industrial-scale funds dedicated to backing early-stage startups with $500k rather than $5 million. They knew the venture math that if only 50 companies / year are sold North of $100 million the entry price for their investments mattered. These funds were active back in 2006 when I was raising money for my second company. As were individuals like Jeff Clavier with SoftTech VC who was also way ahead of the market in spotting this trend.

More recently great funds like IA Ventures, Floodgate, Rincon Ventures, Founder Collective, Freestyle Capital and others have raised money to focus on early-stage investing as a strategy. And many more individuals that I respect are switching from investing as individuals to fund structures to invest in this category like Aydin Senkut (Felicis Ventures), John Frankel (ff Venture Capital), Manu Kumar (K9 Ventures), Chris Sacca (lowercase capital), Dave McClure (500 Startups) and many more.

I have called the creation of Micro VC as the most important change in our industry and I believe it. These people understand that the nature of startups have changed. They have increased the number of investments, they understand that outdated board meeting formats are too slow & unresponsive, they have designed founder-friendly term sheets that can be executed cheaply and they are allowing for a massive increase in the rate of new startup innovation. At least in the consumer & business web.

The larger ones also do more to hold CEO summits, create recruiting databases, set up email distribution lists, create pools of stock options that can be shared across companies, etc.

I still think it was Amazon that created this category not the other way around. Where open-source computing gave us a 90% reduction in our software, Amazon gave us a 90% reduction in our total operating costs. Amazon allowed 22-year-old tech developers to launch companies without even raising capital. Amazon sped up the pace of innovation because in addition to not having to raise capital to start I also didn’t need to wait for hosting to be set up, servers to arrive, software to be provisioned.

Amazon.

I know I’m going on-and-on. I’m not a shareholder. I’m just in awe of what they’ve enabled and baffled that the media doesn’t give this more focus.

In the next post I explore how the changes initiated by Amazon and then propagated by Micro VCs has led to a blurring of the lines in which stages VCs & later-stage investment firms traditionally invest and why this is driving up valuations in private companies beyond common sense.

  • Anonymous

    Mark,

    Agree with you on the stance that we can just use cheap boxes and replace them at well. But we also need the software that manages these boxes and maintains redundant data. This is where the cloud emerged to fill the void via AMAZON. 

    Resulting in many companies coming providing private clouds: rack space, HP solutions, Brocade to name a few. But the main issue in providing these clouds is that most of them still relay on the old datacenter model of: Server to network to NAS/SAN as the underlying infrastructure. 

    Since the model of the cloud depends heavily on Virtual machines via ESX, Hyper-V or Xenserver to optimize for hardware and cost.  We now have thousands of boxes that can compute but the data they need to compute lays in a SAN/NAS which is accessible through a network. 

    The SAN/NAS was a great model when there werent large amounts of data being called by multiple VMs running on multiple servers. Thus this causes a flood of requests from multiple VM’s on multiple servers to access a NAS/SAN that is normally only connected by one switch. 

    Thus the bottleneck forums resulting in lag regardless of if your NAS/SAN has SSDs or not. Your network can only handle certain amount of request and transmit data. Thus the way to solve this issue is to either eliminate the network and localize the SAN/NAS ala the way GFS has done or invest heavily in fiber which increases companies infrastructure costs and increasing the cost of cloud for the end user. 

    Thus the cloud/private cloud revolution for VC’s will lead to the question not how the infrastructure can handle all these VM’s with the current method of Server ->network -> SAN/NAS. So in my opinion the next step for the cloud era will have to be to a fix for broken infrastructure to handle VM’s better. 

  • http://twitter.com/NickyChips Nik Souris

    Mark, great post and thought provoking on several levels (almost slipped a freudian “levers”).  Amazon clearly made fast food business out of application hosting and to their credit they seized the opportunity from what i can tell by learning from the Exodus and early ASP (app soft) models, which they normalized and applied to empower developers.  One might even argue that the “right sizing” of the VC market a decade ago created the necessity that became the mother of this Amazon invention.  But in all fairness a  couple of big guys (aol, yahoo) started the hosting idea with what many still see as the ultimate app – email.  Dropping costs of computing and storage also added fuel to the fire.  Really interesting stuff for the history e-books.

    I am keen to see parts 2 and 3 of your series here.  Hopefully you talk to the future of software. While I can’t put a number on its days – I see writing on the wall for many types of apps, – especially when a Goog or MSFT can allocate $mart people, money, and distribution in a heartbeat.  Look at what Twitter does – give people and business a place to interact – encourage developers to take the market research and product development risks before making a buy, build or squish move.   To me it shifts software development (esp consumer but also some business) from creating functional value (do something smarter, faster, better) to creating entertainment or behavioural value (eyeballs, data, and analytics) at potentially higher $$$/engineer but far fewer engineers = lower exits.  Will be interesting to see if / how  this cycles another shift/fallout amongst the VCs .  Thanks for your continued contributions and inspiration.

  • http://wikidi.com/ Michal Illich

    Actually Amazon EC2 costs significantly more (per CPU power) than your own hardware. The benefits are elsewhere (scalability, tools), not in the price.

  • http://essaychampions.com/ custom essays

    Interesting post its very nice!

  • http://twitter.com/peter_schlegel Peter V. Schlegel

    Mark,

    As a new reader of your blog, I must say it’s great to find something this consistently valuable.

    On this particular post it strikes me that the ‘all else equal’ assumption that you base your writing on is comparing cost of developing and operating to your own experience (which of course was ages ago in internet years) of paid licenses and hardware. Starting a business with a fairly complex model and a relatively large codebase being developed is still costly. Man-years need to be invested.  And while I love Amazon (yes, we are using it) for removing barriers, I still think that the assumption that all of the sudden it’s easy to build and launch is a bit removed from the reality that we’re seeing. In part for reasons you have mentioned previously: we’re in Copenhagen, Denmark so the dealflow is not Silicon Valley-like. But mainly because it’s still sacrificing a year of income for everyone involved in order to get going – or getting some pre-seed investment. Better than the reality you faced. But not easy.

    Sure – all else equal – it’s a lot cheaper to get going, but I suppose that also means that a lot more start-ups – and a relatively larger share not being sustainable businesses – are launched and thus competing for investor attention. This I believe is largely what led to micro-VCs and the competition-based evaluation schemes like SeedCamp etc. who essentially serve as content curators for later-stage VCs. But they have to get paid for risk and effort, so carves out equity for founders very early without having the ability to go the distance with their portfolio companies. So for founders, it remains a trade-off.

    Seems like I’m whining. Don’t mean to. Love incubators, micro VCs and regular VCs. Love being an entrepreneur. But the notion that it’s easy winds me up a bit. Reminds me of my father and his “when I was a kid” stories.

    Again, thanks for posting interesting, readable and valuable stuff.

  • http://blog.chargebee.com Krish

    Michal,
    With all due respect to your opinion I think what Amazon has fundamentallychanged is less upfront investment – converting CAPEX into OPEX. This has provided us startups with a huge lifeline to get into market with minimum amount of money to test waters before scaling up. 

    I do agree with you that Amazon makes it compelling with the ability to scale easily & some amazing tools that are bundled with it.

  • http://wikidi.com/ Michal Illich

    You are right with that CAPEX/OPEX thing. Unfortunatelly this is quite short-sighted approach. The hardware is cheap nowadays (you get decent server for $2000) and having your own hardware is more cost-effective than Amazon EC2 in just half a year.

  • http://wikidi.com/ Michal Illich

    … and if anyone doesn’t want to pay upfront for the hardware, he can either use leasing or rent a dedicated server. This is is more expensive that paying upfront for your own hardware but still way below Amazon EC2.

  • http://tribe.fm/luis Luis Correa d’Almeida

    While on-demand infrastructure and open source are undoubtedly game changers, there are a couple of things that, albeit more subtle, have in my opinion contributed to this shift in a profound way.

    Methodology. We have today a much better understanding around what the first 6 to 12 months of a startup are about. They are about proving the unified theory that we expect to build on and its underlying premises – which will, if proven, allow us to build a business. This means we focus on minimum viable products used by early adopters/customers that provide feedback.

    Stack. On the other hand, actually building software – specifically for the web – is probably an order of magnitude cheaper that 5 to 8 years ago. The Ruby on Rails stack – for example – allows us to ship product quicker and at lower cost. What took a team of 4 to 5 people to develop in 9 to 12 months in Java now takes 2 people and 3 to 6 months to develop in Rails. This translates to 2 founders being able to develop a product and attract early adopters without the need for investment versus a team that costs half a million a year.

  • http://tribe.fm/luis Luis Correa d’Almeida

    That may be so, but with EC2, I don’t need to worry about any of that. I can simply boot up a machine and I’m in business. It is a huge time investment to set up infrastructure. I need to lease the equipment – which will probably include not only the servers, but switches, LBs, rack mounts, cables, power cords, etc… – worry about delivery lead times, find the right ISP and negotiate a flexible plan where I have the ability to add more machines without having to lease the whole 48 units up front. I’m sure it’s doable, but if you are 2 developers at home trying to get the best product out, this is not what I want to focus on. 

  • http://blog.chargebee.com Krish

    But there is also a huge learning curve involved to maintain the server & everything else associated with it. 

    And most importantly it has leveled the playing field for folks like us – I am working out of an office in Chennai (even with all the improvements over the years) is nowhere near the quality of reliable infrastructure and bandwidth that Amazon provides. 

    I can’t imagine setting it up without huge cost like an enterprise does and be in business or to even dream of building a software that would stay in market dominated by PayPal or Amazon’s payment itself. 

  • http://wikidi.com/ Michal Illich

    That’s interesting. Here in Europe you get decent serverhousing from $150/month/server (and you in US can use Rackspace or others for that too). Reading a contract is not fun, but really doesn’t take more than few hours. And switches, load balancers, mounts, cables, etc. can be provided by your serverhousing too – you don’t need to worry about that.

    And if you order managed server it’s the same – you “boot up a machine and you’re in business”.

  • http://wikidi.com/ Michal Illich

    Actually I think there is higher learning curve with Amazon. If you hire linux developer he already knows all the needed stuff to run a server – how to run Apache, MySQL, partition a disk, etc – because it’s the same for his desktop and the server. And experienced linux administrator knows much more (raid, caching, load balancing, networking, etc.). With Amazon you need to learn their API and proprietary tools.

    But it indeed depends on the people around you. I guess in Silicon Valley most developers already know Amazon stack so it’s not a big deal.

  • Craxel

    Mark,

    I think Amazon’s impact with EC2 is even bigger!    There is a paradigm shift in computing that is still to come — where the software running on these utility compute environments is truly horizontally scalable.   The current stacks like LAMP are really vertically scalable and it takes enormous touch labor to scale them and a serious relaxation of consistency requirements to make them work.  Oracle’s stack can’t relax those requirements and it simply doesn’t scale in relation to today’s problems.  Imagine a horizontally scalable software stack (database and application server) that can meet the needed consistency requirements for the enterprise.   We’re working hard on that at Craxel and could not afford to build, test, and demonstrate the algorithms behind it without Amazon EC2 or piles of venture capital.  Someone like us will bring these types of deep technology solutions to market — all due to Amazon.         

  • http://twitter.com/bcwright1217 bcwright1217

    So 10 years ago VCs provided capital to build teams and buy infra. Only the teams have upside:
    1. in what they build
    2. but also the team itself

    e.g. http://blognewcomb.squarespace.com/essays/2010/10/14/cult-creation.html
    >>
    At that time we knew that a talented engineer in a tough to get tech was
    worth about $1.5 million per head.  Thus, I knew with relative
    assurance that since we were going to hire at least 70 people with our
    Series A money, that our worst case scenario was about a $100 million
    exit.
    <<

    The value from number 2 is purely a middle man play – i.e. you're just recruiting talent on your own dime (or the VCs dime) and then cashing in on it later. Middle men serve an important purpose, but they can also be easily disrupted. So the real value comes from the product built and the selling the team itself is more like a safety net.

    But product development doesn't require as big teams now as it used to, this hasn't penetrated to startups yet in my experience, but given the adoption of cloud (when in 2000 building infrastructure was seen as core competency and even one of the ways a company could set itself apart), I'd guess entrepreneurs will rapidly adopt non-staffed approaches to build product (minimum viable team size?) as soon as they mature a bit more.

    Then you go from lean start up, to uber-lean startup.

    I have no idea what the impact on the VC world would be – but at least within companies that are primarily focused on building digital product – I think it will be pretty huge. Why? If you can lean-ify the team building and the infra spend, you push down capital requirements below a certain point access to capital is either easy or not an issue… does this drive the VCs into investing in later stages (where you still need capital to build teams for sales, support?) or into capital intensive areas related to physical sciences?

    I dunno – but I'm pretty sure Amazon is just step one in a two step shake up.

  • http://www.justanentrrepreneur.com Philip Sugar

    Michal, I agree with you.

    Not diminishing Amazon in any way, but the more layers of abstraction you go up, the more dependent you are, on that provider.  Not necessarily a good or a  bad thing.  None of us would consider building our own operating system, but it can be an issue.  You make whatever choice you want, but I agree with you that you do have a choice.

    I think the biggest difference is you now can rent a full rack (in the same datacenter Amazon uses) for $1,500 a month with electric and bandwidth, and be connected to the world at incredible speed.  That is insane!

    You can fill it with inexpensive servers, networking equipment, and load balancers, and for less than $100k have a hell of a resource.  (Or you can spread that out and pay $20k to get going and add a $2k server each month for three years)  That’s the cost of one employee for one year, and the rent of a tiny office. 

    That is a game changer.

    Either way you are taking the technology price right out of the start-up.  It comes down to people and customer acquisition costs.

  • Johnc

    There was a talk by Basil Peters at the Angel Capital Association in Boston last December where he talked about the changes that are occuring in the VC industry. He says that most startup exits are under $30 million, and that this doesn’t work for VCs but works very well for angels.  He gave a bunch of examples like , Flickr, Adsense, Blogger, del.icio.us and others that were all bought by Google or Yahoo for under $30 million. He says that the big companies know how to grow a business from $30 million to over $200 million but they are not very good at creating a new business. He also mentions that it is pretty easy for a large company manager to buy a campany in the $30 million dollar range, but that any purchase over $100 million would require board level approval and a lot of hassle for the purchasing manager and in many casses not worth his time and effort. He mentions that many large company managers see themselves as competitors to VCs since they would rather buy the company relatively cheaply before VCs invest, because they think that the VCs do not add a lot of value. You can view a video of the talk or look at the presentation slides on his site basilpeters dot com

  • Johnc

    Here is a link to a video of the presentation

  • http://twitter.com/DentalDeparture Dental Departures

    Mark:

    I am glad to see you call to attention the unsung hero of cloud computing:  Amazon.   We looked at various cloud offerings when we started our beta and AWS/EC2 won hands down.

    Microsoft has their baby toe in the game with their Azure offering but it faces some serious challenges.  First, Azure locks you into the MSFT stack vs. AMZN which lets you build on multiple platforms .  Second, AMZN keeps lowering their prices for computing and storage at an unheard of pace.  Third, if you know anyone at MSFT you know they are wrestling internally between developing desktop vs cloud solutions which ultimately slows down their rate of innovation.

    As usual you fucking rock!

    Paul

  • Christophe Abiragi

    Mark:Relatively new reader –  led here by a podcast of your talk for the Stanford ETL series.   Found this post interesting and thought I’d share my experiences with Amazon’s AWS. I run a media monitoring company based here in LA.   We are a service business that uses researchers to  gather and filter information to create daily news reports for corporate communications and corporate development teams within companies.    As we’ve grown and worked with larger companies our reports have naturally grown as well.   Our answer has been until recently to add people to the problem, often farming out lengthy and tedious formatting work to lower cost workers around the world.   One of our researchers with an interest in programming pitched me on the idea of building software to automate as much of our process as possible.   Building off Amazon’s platform the risk was very low so it was a relatively easy decision.   As we are not a technology company by design, if the same proposition had been made and required a large upfront investment in the face of total uncertainty my answer would have been different.   The benefits we saw from building out this technology were almost immediate.   Internally we were able to cut our work process significantly.  As an example –  one report that once took 5 hours a day to create now takes 3.    Despite the obvious benefits of lowering labor costs, this also helps to make sure that we’re delivering a uniform product, makes collaboration much easier, makes the work more enjoyable, and allows us to scale more quickly since we can have fewer people working on more reports and our training process for new researchers is significantly faster.   In addition, seeing the benefits that we’ve gotten from our software, we now have a scalable product ready to introduce into the marketplace that will hopefully generate a second revenue stream for the company. As we’re not the type of company that might normally cross your path (we’re not looking to raise money, and did not set out to build a technology) I thought you might be interested to hear how Amazon’s platform is giving small companies the tools to solve problems in ways that they otherwise never would have tried to tackle, and in the process opening up new opportunities. 

  • http://an.ton.io/blog Antonio Rodriguez

    Mark,

    I love the thesis you are exploring but am not sure I agree 100% for two reasons: 

    First, the move away from the Sun/Oracle/Navisite extortion was not binary— I bet if you charted the cost/MB or cost/MIPS of HW and SW you would have seen a steep decline over the decade of 1999-2009 with EC2 and S3 changing the slope only a bit for the very early phases of a startup’s life. The two real big cost reduction drivers where the LAMP stack as you point out and the excess datacenter cages that were going for a fraction of what they used to cost. In fact, by 2005 (when S3 was in beta and there was no EC2), you could “build out” a datacenter in a quarter rack for $25-50K and run it for about $2K/month— all very manageable expenses even for an angel funded startup.

    Second, the thing that doesn’t compute for me still is labor cost. I don’t see how you can do anything meaningful  without about $1M of capital to hire a team 5-7 person for the requisite 12-18 months. Buffer that a bit and you’ve got seed/A rounds that are $1-1.5M which are definitely not $5M but are still fundamentally not that different when you think that you may need all that runway (and a bit more) to really find product/market fit (and no, I don’t think “pivoting” solves this time on saddle problem).

    Looking forward to the other two parts of this series— especially if you get back to how a $1-1.5M A is so different than a $4-5M one.

  • http://tribe.fm/luis Luis Correa d’Almeida

    To your second point – and as I highlighted in my comment above – I think that one of the important differences is that you no longer need 5 to 7 people and 18 months of development to go to market. Look at hyper-productive frameworks and environments on one side and methodologies that help you nail the proposition early.

  • http://www.justanentrrepreneur.com Philip Sugar

    I agree but think your costs were bit high for running it.   I like your term pivoting does not solve time on the saddle.

    I think though the point is that if you run really lean, you can go 18 months with 5 people for $500k.  Yup that is really lean.

    That is a ten fold factor difference than $5M and means you could fund ten for the price of one.  (and no I hate fail fast).

    But basically you always want the prices of your complements to go down and if they do that causes a big boom in your industry.  When gasoline cost $1/gal in 2000 SUV’s sold like hotcakes.  When mortgages could be obtained by anybody in the mid 2000’s housing boomed.

    No its not different this time, but you combine fear of loss (what do you mean you passed on Zynga) , limited partner’s needing their money go to work, and low technology costs. No wonder there is a  “bubble”.

    I’d only say three things:

    1. Lower costs means if you are serving other industries.  I.e. local businesses like Groupon, your margins are going to get squeezed to hell.  Eventually the people actually delivering, will drive your gross margins down close to your incremental cost of delivery.

    2. You better never confuse your value with what you’ve raised.  A $100M pre doesn’t put a single dime in your pocket as an entrepreneur unless you are taking money off the table.

    3. Funding marginal players mean there will be more losses.  If you look at the charts, exits stay relatively (within a 2X swing) even.  It also means you can have some short term dislocations where a player spends stupid money can cause issues for all participants in that market.

  • http://twitter.com/ameensaafir Ameen Saafir

    I love this line: “A 90% disruption in cost spawns innovation – believe me.”.  Exactly our business model.

  • Anonymous

    Good post.  Having helped to launch the Kindle (WhisperNet was the key differentiation to Sony), I saw the technology savvy throughout the ranks of Amazon.  An alternative title for this is the “Bezos Hedge” – profits slim down in retail, they have AWS to help out.  A truly brilliant strategy.   

    However, Amazon is not the only source of the cost changes you describe.  Faster mobile processors, the establishment of (and therefore separation) of the handset hardware from the operating system, and the allure of profits through ad-based applications all result in increased pressure to change.  (They also result in increased obsolescence if not frequently upgraded/ replaced which calls into question the terminal value of any social network).  

    Kudos to Amazon, but also kudos to Intel, Apple and Google and the millions of programmers who dream.  

    It’s time to also start talking about the buzzkill – carrier data rationing.  Unlimited data plans are over – now what? 

  • http://twitter.com/sumagowda Suma Gowda

    Thanks for the article Mark. Its also knowledgeable resources like you that are lowering the barrier to entry…

  • http://www.aaronklein.com/ Aaron Klein

    I’m equally impressed with what Amazon has accomplished but I definitely see some viable competitors out there. Rackspace Cloud, for one, seems like a pretty mature competitor with great uptime and a solid track record. And on the Microsoft side, I know a number of developers using Azure and raving about it.

    That doesn’t make me any less of an Amazon fanboy, but I think the cloud infrastructure choices are getting better and better for entrepreneurs.

  • Dave W Baldwin

    Very good points with bigger meaning moving forward.

  • Dave W Baldwin

    Love the implied message regarding Strange/Maddening.

    For those looking to be truly disruptive, I think it comes down to big players look at things from a position of comfort and old school.  The time frame looking backward defining old school grows shorter and shorter.  Simply put, somebody sitting with a frame of mind based on what they did in ’09 has a tough time understanding what you can do for ’12 at a low cost.

    That said, the game floor is positioned now for the big companies to purchase for $30m what they need.  The only offset is where progress remains slower due to too much fragmentation which raises risks followed by the inevitable ‘less than what we promised’ but ‘still the best you’ll get and setting the future’.

    Moving thru 2015, I advise everyone to use logic determining what the customer wants and build toward the most fluid system of delivering that desire. 

    Also, pay attention to Phillip Sugar’s comment regarding capital required.

  • Rudy Lacovara

    Regarding the comments about EC2 being more expensive than your own servers, this is true after you reach a certain scale, but the point is that EC2 is much less expensive when you’re just getting started.  You can launch your business on an EC2 server for about $300/month out of pocket with no up front costs. Compare that with buying 3 or 4 servers, a firewall, setting up a proxy/loadballancer on one of the servers, signing a multi-year colocation contract, etc…  

    @Mark, I am baffled by the comment about Amazon not having any competitor’s even today. You might not agree with the approach taken by Microsoft and Google but they both definitely have competing products, plus there are other major competitors in the market.    I personally chose RackSpace’s cloudservers to launch my startup.  -rudy
    @Mark:twitter 

  • Peter Coffee, salesforce.com

    Thanks for acknowledging Force.com and Heroku as application platform options for ISVs; not to cavil, but I’d suggest that it’s too narrow to say that Force.com is confined to the crafting of “applications that integrate with [salesforce.com's] core product.” That integration is useful, but there are a growing number of ISVs building apps on Force.com with no particular ties to our mainline CRM and sales force automation application offerings — and there’s also a growing business of SI companies helping enterprises use Force.com as a cloud-extending layer to add value to existing IT portfolios. It will be an interesting future for all of us.

  • Anonymous

    So true. Love the post. Can’t wait for 2 and 3.

  • http://twitter.com/StackSearch StackSearch

    I absolutely agree with you that Amazon deserves much of the credit for bringing the open cloud, but I disagree with your contention that there are not credible competitors.  I think there are three, Azure, Rackspace, and GoGrid.  As was mentioned in a previous comment, Azure has its problems, and is only useful for developers who utilize the Microsoft stack (a considerable number).  Still, they are priced competitively vs Amazon, and if you pair it with the BizSpark benefits available to startups, it can amount to huge cost savings.  Second, Rackspace has the size, scale, and sophistication of Amazon, and also brings their Fanatical Support.  AWS support is not that great, even if you pay for it.  A dark horse is Go-Grid, which has meaningful scale and good support.  Their main drawback, deserved or not, is a lack of name recognition which causes many to pause when committing the guts of their entire operation.

    More are coming on line soon, and could be in the same league in a matter of a year or so.  IBM is starting to offer a product in this space, as is HP and VMWare.  

  • http://blog.canvera.com/ Dhiraj Kacker

    Hi Mark, as usual great stuff. Just catching up on this series. 
    To add some complementary views from a different environment, here is a perspective on “startup costs” in emerging markets that I blogged about : http://blog.canvera.com/2011/05/29/hidden-costs-of-starting-up-a-business-in-india/  (I worked in SV for 7 years before starting a company ~4 years ago in Banaglore focusing on India)

    In essence AWS and other such services exists on top of a strong physical infrastructure, law enforcement and good institutions. While AWS etc can be accessed from any part of the world, the underlying infrastructure still makes it tough to do “capital efficient micro experiments” in emerging markets.  It is also probably why we are not seeing the “Angel Bubble” in India. 

  • http://www.BlueWardrobe.com BlueWardrobe

    Great article – thx

  • http://www.whodoyouknowat.com Lee Blaylock

    Really enjoy your blog Mark.  Having been on both sides of the table, and blogging from that perspective is helpful.  Two topics I’d be interested in your opinion is 1) the continued downsizing of the VC industry (in terms of the #s of funds) and the impact it will have on funds that are presently allocating, but won’t be able to raise another fund (or have a really hard time) and how that impacts their attention/perspective.  I ask that from an LP perspective.  From an entrepreneur’s perspective, what if you are about to get a term sheet from a VC and they aren’t taking the full round and want to bring in other VCs they syndicate with, versus bringing in one I have cultivated and make the intro.  This assumes that both are respected VCs in the same “tier” in terms of industry respect and 1 is willing to let the other set the terms, or at least lead.  I did this successfully in my first deal, but wonder whether it was just good luck they ended up to be productive joint investors.   There are both pros and cons to that approach and I’d be interested in your thoughts.