How to Thrive in a Post AWS World

I recently read an interview by the CEO of Rackspace (now former CEO), in which he said that their company was motivated to innovate in, “a post AWS world.”   The language really struck me. 

What he articulated in the article was that AWS had changed things forever by winning the cloud game - now it was time to work within that reality. Serving AWS (and Azure or GCE) is certainly a profitable business, and doesn’t involve billions of dollars of datacenter investments. It’s most likely a good move for their profits over the next few years, so I can’t fault the logic.

David and Goliath: The Age Old Story?

That said, due to my David and Goliath spirit (go, Dave, go!), I had a different reaction: AWS’s streak of cloud dominance is built on riding the last wave. The next one is certainly coming, and it is coming fast. This was just another round in a long, cyclical battle. So, what does next look like? Who will define it? 

More to the point: how will Packet and other ‘barbarians at the gate’ thrive in a world where AWS is on top, but is also a slow(er) moving giant too busy driving trucks of data into US-East to capitalize on the next big revolution?

Why You Can Have the Advantages and Still Not Win

As the CEO of Rackspace shows us, the current spirit and identity of the cloud is wrapped up with Amazon - a consumer retailer at heart, but with wide ambitions now being funded (in large part) through the profits of its AWS division.  Ten years ago we would have laughed at the thought of Amazon printing billions of dollars of profit from servers.  What gives?

The list of companies that should have dominated "Cloud 1.0" is long. From IT goliaths like Dell, Cisco, VMware, Citrix, and HP to services companies like IBM, Rackspace and the telecoms (Verizon, Century Link, AT&T).  After all, they had the advantages in their court: mountains of capital, decades of related technical expertise, access to limitless hardware, copious broadband / last mile connectivity, etc.  

But at the tail end of Cloud 1.0, the only traditional player in the cloud game seems to be Microsoft, with a distant third place showing from Google.

My read on it is that the traditional vendors approached "Cloud 1.0" by packaging up (very good) products based upon a buyer experience they knew like the back of their hands: incentivized sales teams, long term contracts with complex commitments, and dense feature sheets available only after you sign up for a webinar.  

But that approach pratically ensured they could not win.  Because, the buyer had changed, and they missed it (or didn't want to believe it).  In short, the entrenched players built a product and marketed it to the wrong buyer, letting AWS create the future and a market-leading position with virtually no competition for years.

It's About the Users, Dummy!

I’m a marketing guy, so of course I will argue that marketing was the primary reason why the companies that should have created the cloud revolution (hardware makers, IT vendors, and datacenter providers) missed the boat, while Amazon raided the henhouse.

And by marketing, I don't mean advertising.  That's just a single and over-valued component of marketing.  The primary purpose of marketing is to understand the customer, and to align all the things (product, pricing, experience) behind that.  Hopefully, your core values as a company are flexible enough to handle a changing customer, and you can easily return to guiding principles even when a customer or market changes.

Every group, subculture, and generation has a distinct way of communicating - its own language, dialect, or accent.  And that's why you need to look at the language of the cloud to understand where it came from and how to catch the wave that is coming next.   

Buzzwords, not Brands

Growing up in California, things were not always “cool.” They were “radical” or even “gnarly” before transforming to “intense.” For my parents, it was probably more  “groovy” or “far out.” You get the point.   In startups, and especially tech, trendy words come and go with dizzying frequency, mirroring the rapid pace of innovation. How better to communicate about what’s next than with fresh language?   

When it comes to infrastructure, we’re no different: this year the hot trend is FaaS (Functions as a Service) followed by the enterprise mystery meat of hyper-converged.  Before this, we cycled through an alphabet soup of acronyms like PaaS and IaaS - and don’t even get me started on storage, or the huge alphabet soup of “speeds and feeds” used to show value (SSDs! FiberChannel! FPGA’s! Next-Gen-XYZ-Widget!).  

Hyphens and acronyms abound.

These are all marketing terms, which were created in response to real innovations and real trends. They were supposed to help a fast-growing part of the IT industry (e.g. infrastructure) translate what made it valuable to its customers at various moments in time.

The problem is that they don't really work beyond a single product cycle. Instead, they fizzle into buzzwords, often alienating new customers in the process.  

Why?  Again, because the buyer has changed.  And the new buyer cares about a consumption experience above all else.  They most likely have never taken apart their laptop, or wondered when they would finally get m.2 flash or NVMe in their Macbook.  And that’s why “cloud” now means “AWS style cloud” to so many people, including (especially) the decisions makers that control hundreds of billions of dollars of IT spend.  

Don’t get me wrong - there are still a ton of people who care about SSD’s and even something as obtuse as hyper-converged. At Packet, we are massive fans of the innovation happening in the hardware layer - it’s mind blowing. This stuff solves real world problems. 

But the people who buy speeds and feeds are what I call “IT buyers” and therein lies the fundamental problem: they aren’t making any more “IT buyers”!

Instead, we have a new customer: a developer or (more likely) the software that a developer writes. And while for a long time that consumer didn’t control the purchasing decision, it’s clear that billion dollar businesses will live or die based upon selling to developers.

Transforming Cloud from a Set of Features, to An Experience

On the opposite side, Amazon created an experience targeted at the emerging developer customer.  They could sign up with a gmail address and credit card, deploy via API, pay only for what they used, and never talk to a sales person.  Revolutionary in I.T., but not exactly brain surgery to figure out.  That's why the arguments of "but it's more expensive" or "it's not as secure" or "you can't customize it" didn't stick.  

To this new buyer, it wasn't a detailed value proposition they were after: it was a simple consumption experience that was completely aligned to their tastes, and is now completely non-negotiable.  If it can't be automated, it's antiquated.

By catching a wave of software defined everything and empowering developers, Amazon took ownership of the “cloud” experience for a generation.  And with innovations like Lambda, they are pushing the same strategy of abstraction to the next level.  

So Why Are We Still Talking Hyperconverged?

Most of our industry is still creating B2B language for a product that has moved decidedly toward B2C.  Actually, the more time I spend in the trenches of the cloud, the more I catch myself doing it.  I shudder when I find myself happily talking to customers about "bare metal servers", forgetting that they couldn't care less about servers. 

So why are we (infrastructure folks) doomed to repeat the same mistakes over and over again?  I think it’s because of the very real gap between hardware innovation and the end consumer, which, very quickly, got a LOT closer to the hardware than anyone expected.    

I mean, we used to have a nice fat cushion of Dell’s, HP’s, and IBM’s that would figure out how to turn hardware into product. Now a SaaS startup expects to leverage hardware at scale via an API, and they don't even want to talk about how cool it is!  There are no steak dinners, no 7-year buying cycles tied to budgeted depreciation, and no room for the very difficult logistics of getting hardware designed, built, shipped, connected, and made available to consumers.  That's all plumbing, and nobody likes to think about plumbing.

As a result, we’re still creating marketing that speaks to the world that we know, in which features that cost a billion dollars to develop still matter to our customers.   The truth is: the value matters, but the details of it...not so much.  That's our job.  Just like the impossible logistics of getting an Uber in any city around the world in a few minutes doesn't matter.  It just happens.  

What’s Next and How to Win

To win the next round, we'll need more than just the assets (5G! SoC's! Opteron!) and advantages (capital, edge locations, etc).  We'll need to understand who we are selling to, and craft a ridiculously awesome experience that they want to buy, and recommend to their friends.  We need to capture the imagination of the next generation of software.

There isn't a clear playbook, but I have five suggestions drawn from my background in marketing:

• Look to the Future - Don’t look at today, because that ship has sailed. Build for 3-5 years from now, it’ll be here before you know it. In fact, it’s already taking shape.
• Know Your Buyer - They don’t own a home or a car, but they drink cortados and craft beer. They care about craft, values and can’t be sold through steak dinners.   
• Design the Experience - IKEA is famous for setting the price, and then designing the product. For the next wave of the cloud, pin your dreams on the experience and infill with innovation. What does it feel like?  
• Market Your Values - People don’t buy what you make, they buy what you believe. That’s a good thing for infrastructure investments, because things will change so fast that we'll need huge amounts of wiggle room.
• Listen & Share - Your users will tell you the future, if you’re willing to listen.  

So, what do you think?  And while we're at it: how do you think the cloud will feel in 5 years?  

In a followup post, I'll map out how Packet sees things looking, and what we're building towards.   Would love your feedback!