“Google made a network setting change that accidentally broke the connection between the two data center zones in VMware Engine. The virtual machines themselves kept running fine, but nobody could reach them, and there was a risk that some machines might lose the ability to save data properly. This indicates that even managed cloud infrastructure can experience failures in critical shared network components,” said Pareekh Jain, CEO at  EIIRTrend & Pareekh Consulting. Neil Shah, vice president at Counterpoint Research, said the real culprit here is the SDN orchestration control plane, where a routine internal network update or configuration tweak introduced routing failure across multiple zones. “While most of the physical nodes are distributed for exactly this redundancy purpose, they are still tightly coupled to a singular shared orchestration fabric, so if that control plane crashes, then everything comes crashing down, and the physical distributed nodes become irrelevant.” Stretched clusters fall short Although the outage did not bring down virtual machines, the incident undermined the primary reason enterprises deploy stretched clusters.

Over the past two years, AI build plans have driven a sharp escalation in projected data center power demand. One recent assessment1 found that the U.S. disclosed data center development pipeline reached roughly 241 gigawatts by the end of 2025—an increase of about 159% in a single year—illustrating the unprecedented pace at which AI infrastructure demand is expanding. Forecasts from major analysts indicate that total data center power consumption could grow at least 50% by 2027 and potentially as much as 165% by 2030, with AI training and inference responsible for most of the incremental load.2 At this pace, planned AI capacity is growing faster than electric infrastructure can realistically be expanded. In many markets, available land and fiber are not the limiting factors; dependable megawatt delivery is.3 At the facility level, AI hardware is moving standard designs into new ranges. Power densities that once centered around 10–20 kW per rack are being replaced by configurations nearer 40 kW, with dense AI racks pushing toward 85 kW today and credible roadmaps to 200–250 kW per rack by 2030, though we’ve all seen the reports of even larger. These levels do not only affect cooling and white‑space layouts; they materially change the electrical infrastructure required per room and per building, and by extension the strain on local grids. On the power‑system side, constraints are now explicit. Transmission operators and regulators are stating that current generation, interconnection, and build‑out timelines are not sufficient to accommodate another decade of large demand centers in their present form. Analysts tracking AI data center energy demand point to electricity, grid access, and firm capacity as the primary constraints on new builds, with grid bottlenecks and transmission limitations flagged as risks for up to 20% of planned projects.4, 5  At the facility level, AI hardware is moving

The Hidden Constraints of Delivery If power gets the headlines, supply chain and logistics often decide the schedule. Kleyman notes that a seemingly small missing component can delay a multibillion-dollar facility. A busway, switchgear component, cooling element, or logistics failure can ripple through construction sequencing, commissioning, customer handoff, and revenue recognition. “The weakest link may not be the most expensive component,” he says. That reality receives sustained attention across the Summit agenda. Day One’s “Beyond the Dashboard: Active Exception Management for Hyperscale AI” features CargoSense CEO Rich Kilmer in a live case study examining how organizations are moving beyond passive shipment visibility toward active exception management. For hyperscale AI projects, supply chain disruption is not simply about delayed shipments. It can affect site readiness, construction sequencing, commissioning windows, and the ability to bring capacity online as planned. Day Two’s “The Hidden Constraint: Supply Chains in the Age of AI Infrastructure” continues the discussion, examining how global supply chains are becoming a defining constraint and differentiator in AI data center delivery. The execution lens sharpens again on Day Three with “The Last 90 Days: Solving the Final Infrastructure Bottlenecks Before Go-Live.” This session focuses on the phase where projects can be won or lost: generator delivery, electrical integration, controls validation, startup sequencing, fuel systems, utility coordination, commissioning, and operational readiness. Even projects that have secured power, capital, customers, and equipment can face costly delays if the final stretch is not executed with precision. In the AI infrastructure era, the last 90 days may determine whether a project becomes energized capacity—or another delayed announcement. Capital Meets Execution Reality The Summit also examines whether capital is moving in step with what can actually be built. Day Two’s investment panel, “AI Infrastructure Investment: Bubble, Breakthrough, or Both?” will assess how investors are underwriting risk

Matt Vincent is Editor in Chief of Data Center Frontier, where he leads editorial strategy and coverage focused on the infrastructure powering cloud computing, artificial intelligence, and the digital economy. A veteran B2B technology journalist with more than two decades of experience, Vincent specializes in the intersection of data centers, power, cooling, and emerging AI-era infrastructure. Since assuming the EIC role in 2023, he has helped guide Data Center Frontier’s coverage of the industry’s transition into the gigawatt-scale AI era, with a focus on hyperscale development, behind-the-meter power strategies, liquid cooling architectures, and the evolving energy demands of high-density compute, while working closely with the Digital Infrastructure Group at Endeavor Business Media to expand the brand’s analytical and multimedia footprint. Vincent also hosts The Data Center Frontier Show podcast, where he interviews industry leaders across hyperscale, colocation, utilities, and the data center supply chain to examine the technologies and business models reshaping digital infrastructure. Since its inception he serves as Head of Content for the Data Center Frontier Trends Summit. Before becoming Editor in Chief, he served in multiple senior editorial roles across Endeavor Business Media’s digital infrastructure portfolio, with coverage spanning data centers and hyperscale infrastructure, structured cabling and networking, telecom and datacom, IP physical security, and wireless and Pro AV markets. He began his career in 2005 within PennWell’s Advanced Technology Division and later held senior editorial positions supporting brands such as Cabling Installation & Maintenance, Lightwave Online, Broadband Technology Report, and Smart Buildings Technology. Vincent is a frequent moderator, interviewer, and keynote speaker at industry events including the HPC Forum, where he delivers forward-looking analysis on how AI and high-performance computing are reshaping digital infrastructure. He graduated with honors from Indiana University Bloomington with a B.A. in English Literature and Creative Writing and lives in southern New Hampshire with

Alberta represents the biggest point of contention in Canada’s data center strategy. The province is aggressively pursuing data center development with its Artificial Intelligence Data Center Strategy. It has abundant natural gas, large land parcels, a deregulated power market, experienced energy developers and political leaders actively courting AI infrastructure. That makes it attractive to data center operators that care most about speed to power. Alberta has also promoted “bring your own generation” models, where data center developers pair facilities with dedicated generation rather than relying entirely on the public grid. But Alberta’s electricity system is much more carbon-intensive than Québec, British Columbia, Manitoba or Ontario. The same feature that makes it attractive for development, potential  large AI build-outs powered primarily by natural gas, would undercut Canada’s claim that its data centers can run on some of the cleanest power in the world. Saskatchewan illustrates another version of the opportunity. Bell Canada’s planned 300-megawatt AI data center in the Rural Municipality of Sherwood near Regina is a major signal that large-scale AI infrastructure can move beyond the traditional Toronto-Montreal-Calgary corridor. The project combines domestic telecom infrastructure, sovereign compute ambitions, hyperscale tenants, fiber partnerships, Indigenous procurement participation and closed-loop cooling. It also shows why power availability is now the deciding factor in site selection. At 300 megawatts, a single facility becomes a grid-planning event, not merely a real estate development. British Columbia, meanwhile, is trying to prioritize power among competing industrial demands. Data centers are arriving at the same time as mining, LNG, manufacturing, forestry, hydrogen and electrification projects. The province has moved toward limiting and screening certain high-load uses, including data centers and cryptocurrency mining, so that scarce clean electricity is allocated to projects with the strongest public benefit. This seems to be a preview of the future for these industrial

For much of the past two years, the data center industry’s central question was whether artificial intelligence demand could sustain the unprecedented scale of infrastructure being announced in its name. That is no longer the most urgent question. The more immediate concern is whether developers can assemble the power, land, cooling systems, interconnections, capital, political support and community acceptance required to convert that demand into operating capacity. “AI infrastructure has moved from being a fast-growing demand segment to becoming the organizing principle of data center development,” said Colby Cox, Managing Director for the Americas at DC Byte, during a recent appearance on the Data Center Frontier Show podcast. The phrase captures a fundamental shift. AI is no longer simply one workload category competing for space inside the conventional data center market. It is beginning to determine where facilities are built, how campuses are financed, how electrical and mechanical systems are designed, and which regions can realistically participate in the next phase of digital infrastructure growth. The market has also moved beyond incremental expansion. Campuses planned around hundreds of megawatts—and increasingly multiple gigawatts—are no longer treated strictly as outliers. They are being conceived from the beginning around GPU density, liquid cooling, accelerated deployment and enormous concentrations of electrical load. Behind that transformation lies a second, increasingly decisive reality: Demand may be abundant, but deployable power is not. Executable Power Separates Announcements From Infrastructure From DC Byte’s market-intelligence vantage point, the dividing line between announced capacity and capacity likely to reach operation is increasingly straightforward. Has the power been secured? When can it be energized? Can the grid—or an onsite alternative—support the project’s intended phasing? “The market is no longer constrained primarily by demand or capital,” Cox said. “It is constrained by executable power.” That distinction matters because a headline capacity figure