The Data Center Industry is in Crisis: And Now What?

Over the last 2 yearsData Center Industry, on like-for-like data center deals with the same provider and renewals without growth, we’ve frequently seen price cuts of 8%-25% depending on location and buyer leverage.  Zahl Limbuwala, recently stirred the pot in his Open Letter to Data Center Investors. To sum it up very briefly, he sees deep structural problems for anyone looking to invest in the data center industry. Problems that he believes many owners and investors are complacent or in denial about. The concerns he highlights are myriad: facilities that are outdated long before depreciating; lower competitiveness with emerging cloud options; easier migration and reduced lock-in; and the unlikeliness of becoming a supplier to the cloud.

The reaction I got on polling the numerous visionaries within and outside of RampRate is mixed.

  • Zahl is right in many ways. And if anything, too gentle on investors who still see a data center as a real estate play. It’s a technology investment, and winners and losers are determined by power configuration / density, airflow engineering, value added services, integration with cloud providers to optimize compute as a unified stack, telecom, etc. Not to mention the uncertainties of future demand and the changing buyer mindset that makes them less complacent and pliable even when they do stay in existing solutions.
  • No market moves as fast as visionary types think. My team is still negotiating renewals for mainframe contracts right alongside cloud deals for the same clients. It’s not as though what was stopping migration to newer technologies was a lack of confidence and with transparency; the floodgates will open. There’s a complex migration path and a lot of tradeoffs that can make co-lo the right place to put some workloads for years to come. Plus, demand is growing quickly enough for buyers to forgive a lot of warts.
  • It’s thrilling to deliver a prescription – an actionable recommendation for the industry.  Of course, there’s an implicit recommendation – optimize your mess for less to stay afloat as others sink. But less facile explanations have to consider the many types of data center investors that are out there – whether enterprises serving their internal constituencies or providers to others; with sunk costs or just breaking ground; technologists; real estate owners; venture finance.
  • Bruce Taylor, EVP, North America for Datacenter Dynamics, puts it quite eloquently—“The nature of designing and delivering data centers is already changing dramatically in a number of directions. Global mega-scale… and modular. Regional mid-scale… and modular. Local micro-scale… and modular or container. These are the trends with modularity becoming an answer to overbuilding capacity and over-provisioning; pacing build-out to more nearly match business demand.

Why it’s Worse than You Think

Deeps truths are resonating through the world.  And in some ways, it’s worse than  even Zahl’s doom-and-gloom prognostications:

  • It’s not even close to a property business. The idea of “sweating their asset” is now beyond wrong – it’s delusional. Most of the cost – and the value – of the property is tied to power. Without a good power contract, equipment, and airflow design to get a solid PUE and watts per square foot, you’re dead in the water. And with the much more frequent equipment refreshes, your investment in these rises as well. And there’s more. With rising hybrids of co-lo and cloud strategies, telecom density becomes a bigger lever and expense. Value added services and support staff add to the pie. At the end of the day, the real estate cost becomes a tiny fraction of the pie, and making a great real estate bet counts for nothing if the other parts are suboptimal.
  • Everyone is in the dark about supply / demand. Every attempt to accurately forecast future tradeoffs of supply and demand has flopped. T1R / 451 Group is easy to pick on because they were so public about their forecasts about upcoming supply shortages and increasing prices for the last 5 years. But no one – not buyers, not sellers, not real brokers – truly grasped the impact of everybody doing more with less by turning their dedicated servers into hypervisors. And I’d be lying if I told you I had a crystal ball for 2-3 years from now. For the next 6 months, sure – rates are not going to increase a lot because long-term contracts still come with a discount and not a surcharge as they did before the last run-up. But making a 20 year investment when you don’t know if the market will still be there when you are ready to open the doors. That’s tough.
  • The cloud is more than just “not your friend” – it breaks the business model by reducing stickiness. The co-lo (and managed hosting) business model is founded on inertia. Host a startup in its early years, and it will stick with you as it becomes a sensation. Host a low-latency trading app for a bank and soon enough someone will start dropping e-mail servers in because it’s convenient. The cloud mindset is different – instead of satisfying many workloads with a decent solution, buyers now look for the right place to put each workload, with as little commitment as possible in case that right place changes. And that mindset creates a drive for optimization that kills the “easy money” legacy deployments, whether it’s 5 and 10-year contracts, 1 megawatt customers paying same rates as they did at 50kw, non-coterminous contract ends, etc.

Why Things Are Not That Bad

However, as much as the future is murky for data center investors in the abstract long term, in the tangible and immediate future, things are by no means doom and gloom:

Data Center Industry

  • First, the pace of business is more tectonic than the visionary types think. For every customer using ServiceMesh to orchestrate movement of workloads between clouds, there’s more than one that needs a solid support deal for his mainframe. Sometimes they’re the same customer. Which is to say that you don’t have to be on the cutting edge to have a market. And just as IBM continues to cart in wheelbarrows of money from mainframes, someone will do fine on co-lo when it’s not hot or sexy.
  • Secondly, the reason it’s slower than you hope is because the devil is in the details. If it weren’t, we wouldn’t have much of a business in cloud migration. But as it is there are very real and complex challenges to using cloud services that we work on every day – measuring the requirements and interplay of the different workloads in the business, figuring out the right charge model, location / latency. And at the end of that analysis, there are places where co-lo is better – or at least a better stopgap for the app lifecycle – and nothing is as permanent as a temporary stopgap. It’s not as though the cloud is new – or that everyone was waiting for someone to say it’s easy to move. Most of your mess isn’t moving to the cloud for years anyway.
  • Thirdly, as much as there is uncertainty about technology being able to absorb future demand, pretty much everyone agrees there’s lots of demand coming. The wave of virtualization and energy-efficient servers that froze new deployments for the last several years has probably peaked; big data continues to get bigger; the internet of things, or even internet of everything is coming. In that wave of demand, many sins will be forgiven. Nothing will save the worst offenders like the carriers with 60 watts / square foot capacity, but many mid-tier providers that aren’t anyone’s first choice can survive with a bit of adaptability.

So What Now?

So we established there’s a crisis mounting for the data center and colocation industry. Maybe not as quickly as some prognosticators would have you believe. But when it comes, it will hit hard. The real question is what to do about it. And that depends on what kind of data center investor you are:

  • Enterprise Owner / Operators that serve their internal constituencies need to make long term planning decisions around expanding, modernizing, repurposing, or even selling their in-house facilities. And those decisions should come in the context of as much data as possible:
    • An assessment of the current costs of the entire stack for each workload – a bill of IT that can be provided on an ongoing basis by a TBM solution such as Apptio.
      • A comparison of these costs to the various solutions in the market ranging from co-lo to managed hosting to cloud, which is part of RampRate’s core practice for assessing cost of compute.
      • A detailed strategy for allocating workloads flexibly between the solution platforms, with multiple standing contracts for on-demand capacity, and possibly a cloud orchestration platform like CSC / ServiceMesh’s in place.
  • Providers / Co-los with sunk costs and older facilities need to identify the best way to make their existing investments play in the new world to come without throwing good money after bad. Options include:
    • Negotiating better deals with key suppliers such as electrical utilities and partners such as telecom suppliers and optimizing power efficiency to squeeze every penny of cost savings out. This is where power optimization now only saves cost, yet helps tailor the right use.
    • Understanding the interplay with cloud solutions and joining in with them as Equinix did with its Cloud Exchange to allow customers to hybridize their platforms – rather than attempting for lock-in by making migration harder.
    • Working with RampRate as a channel partner to find the buyers and workloads for whom your solution is the absolute best fit.Knowing when to cut your losses and divest a facility that’s not going to be in the top 75% in the short term or the top 25%-50% in the long term.

      Data Center Industry

      Tools like RampRate’s workload TCO modeling analytics help enterprises understand the tradeoffs between various hosting placement options, payment and risk scenarios.

  • New Incoming Providers that are looking to make new wise investments should think carefully about their levers of success before breaking ground:
    • Is the power contract long term, flexible, and competitive?
    • Is there enough fiber coming into the building (not wish casting who will want to come in when your customers are already there)?
    • Is there something that I’m absolutely going to be best at? PUE / green power? Density? Network? Service / support? Latency to some key location like a stock exchange? This is not a world for also-rans.
    • Is everything I build modular? How quickly can I expand so that I don’t have to run idle capacity?
  • Outside Investors that are not necessarily experienced in the data center market should doubt everything and everyone before signing the check:
    • The business models of the past are almost guaranteed to not work – the new business plan should have not just the economics of the last few years, but of the next big thing.
    • Middle-of-the-road generalists are not worth investing in with an age of specialization and optimization coming. If someone can’t say what they’re best at, they won’t succeed.
  • Cloud / Managed Hosting Capacity Owners that are adding many layers of services on top of the data center should pay close attention to what happens in an atmosphere of commoditization and look to find the vertical or technology stack where again, they can be a leader. A lot of our work with providers is precisely on identifying that angle and customer set with unique value.

The bottom line is that every crisis is an opportunity to evolve and differentiate. In an era of more disruption and disintermediation, you need to keep an eye on the pulse of the market, avoid long-term commitments, and invest in flexibility. And next time we’ll tackle what buyers can do to take advantage of this disruptive time for the providers.

Speak to Us

Leave a Reply