Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesSome links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Data-center innovation is not a single technology upgrade. It is a facilities strategy decision: whether to own and operate, lease space and power, use colocation or cloud services, and how quickly to add capacity without paying for equipment and rooms that will sit idle. A sound plan starts with measured demand, total cost of ownership, required autonomy, and a building’s actual engineering limits.
The phrase “new era” comes from a Building Design + Construction feature published October 10, 2016. Its market forecasts and project figures are historical, not 2026 benchmarks, but its decision framework remains useful for evaluating a new facility or a conversion.
As an Amazon Associate I earn from qualifying purchases.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →What the 2016 “new era” actually described
The article linked data-center change to the growth of information, content, and computing. It used “SMAC”—social media, mobile, analytics, and cloud—to describe demand drivers that were increasing both the volume of workloads and the speed at which organizations needed infrastructure. The practical result was a more complicated choice than simply building a larger server room.
Organizations were weighing three related questions: whether owning and operating a data center was preferable to leasing space and power; whether to build new or retrofit an existing facility; and how much operational and computing autonomy to relinquish or retain. Those questions still need to be answered together because a low construction price can be offset by inflexible capacity, while outsourcing can reduce facility work but limit control over location, security, or operating decisions.
Popular on Amazon
| Figure reported in the October 2016 article | Attribution and qualification |
|---|---|
| More than 6 million U.S. data centers, with 8.6 million forecast by 2017 | IDC forecast as reported in 2016; not a current count |
| 12.1% compound annual growth in multi-tenant colocation centers, 2015–2018 | JLL projection reported in 2016 |
| Cloud-managed services projected to double to $76.7 billion by 2021 | JLL projection reported in 2016 |
| 40% increase in installed servers from 2010–2020, alongside a projected 10%–40% reduction in industry energy use | JLL projections reported in 2016; not a measured industry result |
| Enterprise-owned facilities hosting 71% of enterprise IT assets | Uptime Institute survey finding reported in 2016; not a present-day share |
These numbers show why the article sounded a warning against one-size-fits-all planning. They do not establish what the market looks like in 2026. Current forecasts, utility conditions, equipment performance, and cloud-service terms would have to be checked separately.
Choose the operating model before choosing the building
Facility strategy should follow the workload and business requirement, not the other way around. The four broad models below are decision categories rather than mutually exclusive commitments.
Recommended Free Tools
| Model | What the organization controls | What it gives up or takes on | Best question to answer |
|---|---|---|---|
| Own and operate | Site, building systems, security policy, maintenance practices, and deployment schedule | Capital cost, staffing, power procurement, compliance work, and long-term utilization risk | Do critical workloads justify retaining facility and operational autonomy? |
| Lease space and power | Often the equipment and some operating decisions | Less control over the base building, expansion timing, and landlord constraints | Can a predictable lease meet resilience, security, and growth requirements? |
| Colocation | Typically hardware, network choices, and workload placement | Dependence on a provider’s building, operating procedures, and available expansion capacity | Is shared infrastructure faster or cheaper than delivering a dedicated site? |
| Cloud services | Workload architecture, data policy, and service configuration | Physical-location control and some direct visibility into facility operations | How much operational and computing autonomy can the business safely relinquish? |
| Hybrid | A deliberate mix of the above, often separating critical, regulated, burst, and commodity workloads | More integration, governance, and capacity-planning complexity | Which applications need dedicated control, and which benefit from elastic or outsourced capacity? |
The 2016 article specifically noted that enterprise facilities were not disappearing: its cited Uptime Institute survey found that enterprise-owned centers still hosted 71% of enterprise IT assets. Treat that as a dated observation, not evidence that ownership is preferable today. A hybrid design can be rational when latency, security, data residency, or specialized hardware makes some workloads unsuitable for an external service while other workloads can move elsewhere.
Build new or retrofit an existing facility?
A new build offers a clean structural grid, purpose-designed power and cooling, and a site plan arranged around current requirements. A retrofit can reach service sooner and reuse a building, substation, roof, or utility connection. Neither option is automatically cheaper. Suitability must be demonstrated through engineering and due diligence.
Also popular on Amazon
When a retrofit can make sense
- An existing building has adequate utility access and a path for high-capacity electrical distribution.
- Structure, floor-to-floor height, loading access, and fire protection can support the intended equipment.
- Telecom providers can reach the site through diverse routes.
- The building’s location, flood exposure, seismic conditions, and roof condition fit the resilience requirement.
- Capacity can be added in stages instead of forcing a fully built-out interior on day one.
Retrofit screening checklist
The article’s screening prompts are a starting point, not a substitute for code review, geotechnical work, or stamped design. Test each item before committing to a conversion:
- Floor-load capacity and the need for structural reinforcement.
- Clear ceiling height and space for busway, cable pathways, ductwork, and maintenance access.
- Existing electrical capacity, service entrances, generator location, switchgear space, and access for new equipment.
- Water availability and the feasibility of the selected heat-rejection system.
- Telecom connectivity, carrier diversity, and physically separate entry routes.
- Seismic resistance and the anchorage requirements for racks and mechanical equipment.
- Flood-zone exposure, drainage, finished-floor elevation, and emergency access.
- Roof condition, roof loading, penetrations, and space for cooling or generation equipment.
- Power sourcing, utility reliability, and the availability of additional service in the required time frame.
Brian Johnston of QTS Realty Trust described the retrofit logic this way: “We’re identifying cities where we can step into a retrofit environment and take advantage of an existing building’s infrastructure while mitigating our risk.” The risk is mitigated only when the existing infrastructure is verified; an attractive shell with inadequate power or cooling is not a ready data center.
Stage capacity instead of guessing the final load
Demand forecasts contain peaks, valleys, and timing uncertainty. The article describes completing a core and shell, then installing interior infrastructure as demand or technology warrants. This approach can preserve expansion space and shorten the time between a signed requirement and usable capacity.
- Define the initial critical load. Document IT load, network equipment, storage, support rooms, and the redundancy level required at opening.
- Map the expansion envelope. Reserve structural, electrical, mechanical, and cable-path capacity for the next phases rather than reserving only empty floor area.
- Separate shell work from fit-out work. Complete items that are difficult to add later—utility service, structural reinforcement, fire separations, and major risers—while delaying racks and selected mechanical equipment until demand is real.
- Set trigger points. Tie each expansion phase to contracted demand, measured utilization, lead times, and available power, not to an optimistic long-range number alone.
- Protect maintainability. A staged plan still needs clearances, equipment-removal paths, testing zones, and safe access while adjacent phases are operating.
Staging is not the same as undersizing. If a future electrical room, generator yard, or cooling loop cannot be added without interrupting live operations, the “future” capacity is only theoretical.
Also popular on Amazon
Use total cost of ownership and consistent units
Construction cost alone can hide the cost of power, maintenance, staffing, replacement cycles, financing, downtime exposure, and stranded capacity. Compare ownership, colocation, and cloud alternatives over the same planning horizon and workload boundary.
The 2016 article gives a warning example: a financial-institution facility was described as costing $30,000 per kilowatt-hour, compared with $9,500 per kW for colocation construction, in an anecdotal comparison attributed to David Schirmacher. The units appear differently in the article and are not a validated like-for-like benchmark. Do not silently “correct” them or use the figures to price a project. Before comparing any cost-per-capacity number, confirm whether it covers IT load or critical power, shell or fit-out, land, utility work, financing, and recurring operations.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallA practical evaluation should show, for each option:
- Up-front capital and the timing of each phase.
- Recurring electricity, fuel, water, maintenance, staffing, insurance, taxes, and connectivity costs.
- Expected utilization and the cost of unused capacity.
- Expansion cost and the time required to bring it online.
- Requirements for security, compliance, disaster recovery, and vendor management.
- The financial effect of outages, migration, or an early change in workload.
Cooling and server density are changing the design brief
The article presents cooling, server technology, renewable power, and tolerance for higher operating temperatures as possible ways to reduce energy and water use. It does not provide a quantified, generalizable savings figure for the retrofit example, so those ideas should be treated as design options requiring project-specific modeling.
Also popular on Amazon
The Liebert DSE retrofit example
For a confidential telecom client, Vanderweil Engineers based a project on Emerson’s Liebert DSE pumped-refrigerant economization system. The Building Team added a perimeter-cooled raised floor and removed two of six cooling units. Christopher McLean said, “That was unheard of before.” The account demonstrates that a cooling redesign can change how much installed equipment is needed; it does not promise the same result in another building, climate, load profile, or operating regime.
High-density racks require an explicit heat-removal plan
Robert Haley of HDR reported, “I’ve seen racks up to 100kW at some of these sites.” That is an observation about certain college, research, or military sites, not a market-wide rack-density statistic. If a project expects unusually dense racks, model airflow, liquid or refrigerant options, electrical distribution, structural loading, service access, and failure modes before selecting a room layout.
Free tools Windows power users keep installed
One-click scans. No signup required.
Standardize where it helps; customize where it protects the mission
Repeatable designs, procurement packages, and commissioning procedures can help providers expand rapidly and control cost. Standardization is especially valuable for components that are difficult to redesign safely at every site: electrical rooms, controls sequences, monitoring points, and test scripts.
Flexibility remains essential for institutions with distinctive IT, security, research, military, or people-flow requirements. A college laboratory may need equipment movement and specialized environmental control; a military installation may require restricted circulation and security zones. A template that ignores those constraints can create operating risk even when it reduces construction variation.
Also popular on Amazon
Vanderweil’s McLean summarized the balance as: “What used to be maximally modular is now minimally modular for maximum flexibility.” In practice, that means standardizing the repeatable backbone while leaving room for client-specific rooms, security boundaries, equipment types, and expansion sequencing.
Historical projects that illustrate the strategy
| Project example reported in 2016 | Details reported | Lesson for planners |
|---|---|---|
| QTS Piscataway | 360,000 square feet and 18 MW, with a stated plan to add 8 MW and exceed 26 MW of critical power | Plan the shell, utility path, and expansion sequence together; the stated figures are project-specific and historical. |
| T5 Elk Grove Village | 208,000 square feet, with an adjacent four-acre lot described as capable of supporting another 20 MW | Adjacent land can preserve a future power and building option, but capability still requires utility and engineering verification. |
These examples are not current specifications or guarantees. They show why site selection should include room for expansion, not just the first phase’s square footage.
A decision path for a 2026 project using 2016 lessons
- Classify workloads. Separate latency-sensitive, regulated, security-sensitive, burst, and ordinary workloads. Identify which can move and which require dedicated control.
- Quantify demand. Record present IT load, peak load, growth timing, rack-density ranges, and confidence intervals. Avoid converting a distant forecast directly into installed equipment.
- Compare operating models. Price owned, leased, colocated, cloud, and hybrid options on the same workload, resilience, and time horizon.
- Screen sites before designing rooms. Verify structure, power, water, telecom, seismic, flood, roof, access, and utility expansion conditions.
- Choose new build or retrofit. Select the option that meets the required schedule and expansion plan after due diligence, not the option with the lowest initial shell estimate.
- Design a staged backbone. Install hard-to-add infrastructure early, reserve expansion paths, and define measurable triggers for each fit-out phase.
- Model cooling and density. Test the chosen cooling approach against actual equipment, climate, water constraints, maintenance needs, and failure scenarios.
- Set the standardization boundary. Use repeatable details where requirements are common and preserve design latitude where mission, security, or workflow demands it.
- Revalidate with current evidence. Because the cited article is from 2016, replace its market forecasts and historical project assumptions with current utility data, vendor proposals, codes, and independent financial analysis before approval.
Bottom line
The innovation described in the 2016 feature was fundamentally a delivery model: match capacity, ownership, cooling, and construction timing to the business instead of treating a data center as a permanently fixed building. New construction, retrofit, colocation, cloud, and hybrid strategies can all be appropriate. The defensible choice is the one supported by measured demand, consistent cost definitions, verified site conditions, staged expansion, and exactly as much operational autonomy as the mission requires.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
