DCD Connect NYC 2026 — Session Guide

Delegates collect their badges and get their personal agenda and app questions answered

The Show Floor opens and all partner Stands and Lounges are open for business!

Agenda 8:30am - 9am – Arrival/check in 9:00am - 9:15am – Welcome and Opening Context & Industry Inflection Point9:20am - 9:45am – iCA Fireside with Bloomberg9:50 - 10:05am – The Road Ahead: iMasons as a Global Institution10:10 - 10:45am – Member Working Session10:45 - 11:20am – Table Topic Report Out11:20 - 11:30am – CEO Reflection, Synthesis & ClosingThis Co-located Member Summit, hosted by iMasons, will be held prior to the official opening of DCD>Connect | New York. Join the iMasons Member Community to discuss the latest topics within the digital infrastructure industry.

This taster session introduces the core principles of data center design, providing a high-level overview of how facilities are planned, designed, built, and commissioned, from concept through to operation.The session will also touch on data center design in the age of AI, examining how emerging workloads are influencing power density, cooling strategies, resilience planning, and future-proofing considerations. Designed for professionals from across the industry, this session supports onboarding, cross-functional collaboration, career development, and customer-facing roles. No prior technical background is required.The taster session includes a 45-minute overview, interactive Q&A, and valuable networking opportunities, open to all attendees of DCD Connect New York.For more details on the course, visit the Data Center Design course page.

Operators are racing to deploy AI, triggering the fastest, largest period of market expansion the digital infrastructure industry has ever seen. But building at scale and pace is also testing limits, with constrained grids, rising costs and increasing public scrutiny putting additional pressure on data centers to demonstrate they can sustain growth. The challenge is no longer about whether we can build capacity, it is about whether the industry can deliver it responsibly, efficiently and with long-term resilience in mind.

In today’s technology landscape, AI and HPC are becoming ever more interconnected, making supercomputing the optimal proving ground for the most innovative use cases for these different modes of computing being used in concert. From Frontier, to the recently announced Lux and Discovery Supercomputers, ORNL is proving the potential of these powerhouses to move beyond performance boundaries whilst prioritizing sustainable computing. Join us for our opening keynote talk with Scott Atchley, Chief Technology of Oak Ridge National Laboratory, as he offers insights into the future trajectory and alignment of AI, HPC, and Quantum computing.

The GPU Cloud market is experiencing rapid growth, fueled by significant investment and new entrants. However, mature providers face the challenge of balancing this investment against market pricing, all while trying to keep pace with innovation. Although the demand for AI services appears to be on a steep upward trajectory, many newer industry players are currently limited to offering only Bare Metal (BM) as a Service. The crucial shift is moving from selling foundational BM services to offering higher-value, specialized services. This keynote will explore the current state of the market and what is required to achieve this necessary transition.

AI workloads are putting immense pressure on traditional data center power systems, pushing the limits of backup, distribution and on-site generation technologies. As densities continue to rise and power consumption profiles evolve, operators must decide whether incremental upgrades can meet the new era of power demand, or if the time has come for a complete redesign of data center power infrastructure.

As AI workloads reshape the scale and urgency of data center development, the industry needs metrics that look beyond traditional notions of facility efficiency. Building on The Green Grid’s recent work—including the Data Center Resource Effectiveness (DCRE) metric, the Water Usage Impact (WUI) metric, and the first release of the IT Work Capacity (ITWC v1)—TGG is now developing the next critical benchmark: the Energy Usage Impact (EUI) metric.EUI introduces a power grid‑aware, system‑level perspective—evaluating how AI data centers affect power grids, transmission and distribution networks, as well as the growing role of behind‑the‑meter power generation.At DCD Connect NY, this session offers operators, designers, and policymakers an early look at the EUI framework and its implications for siting, scaling, and grid‑aligned design of high‑density AI facilities. Attendees will have the opportunity to directly influence the metric’s development, with an open call for contributors and prospective members—supported by TGG’s expanding collaborations with organizations such as ASHRAE and ISO/IEC.Join us to help shape the next generation of sustainability and reliability metrics for the world’s energy‑intensive digital infrastructure.

Liquid cooling has moved from niche to necessity as AI and high-density workloads push thermal limits beyond what air alone can handle. Yet, the industry remains divided: should operators double down on single-phase and direct-to-chip systems now, or wait for two-phase technologies to mature? This debate tackles where the tipping points are for different liquid cooling technologies, and how operators are adapting their thermal strategies in the face of rapid workload evolution.This debate will address:How to define the ‘right’ approach to cooling for your operationWhen will single-phase and cold plate technologies will give way to immersionHow standards, materials and fluid safety considerations are evolvingThe operational and commercial factors driving long-term cooling choices, and investment decisions today

The data center industry is under pressure from every direction. Surging demand is driving developers, EPCs, and investors to deploy capital at unprecedented speed, yet the race to build is colliding with a hard reality: labor shortages, supply chain complexity, rising costs, and evolving community perspectives. From land acquisition to energization, delivery has become as much a people problem as a materials and technology problem.The materials and systems that make up a data center, from power cabling to data infrastructure, carry real consequences for fire safety, emissions, and long-term community impact. Developers who can translate those decisions into clear, credible sustainability and compliance claims find it easier to bring municipalities and communities along. This debate explores how the industry can meet record demand, solve the delivery crisis, and build the public confidence needed to keep building. The question is no longer whether social license matters, but how to earn it at scale.This debate will consider:How developers are managing the full delivery chain, from procurement to operations, amid labor and resource constraintsWhether AI is solving the workforce problem or deepening the confusion around itHow material and construction choices shape community safety perceptionsBuilding trust through transparency, sustainability credentials, and consistent engagementWhat a credible, industry-wide social impact framework could look like

Data centers are facing an acute shortage of talent in mission critical roles, just as AI reshapes how facilities need to be monitored, maintained and optimized. Some argue these technologies will help close the current skills gap while others believe the introduction of AI-aided operations will lead to greater hurdles in training, developing hands-on experience, and operational resilience. This debate explores whether AI will narrow the divide between human and machine capability, or make it harder to close.In particular we address:Has AI had any marked impact on the workforce yet?Whether AI could shift - or shrink - traditional skills requirementsCould AI help close the gap for scarce expertise to drive operational resiliency?How to manage the risk of overdependence of AI and automated systemsThe need to drive AI-literacy in the workforce

The evolution toward agentic AI systems and reasoning-baed models has revealed a fundamental architectural chokepoint: GPU high-bandwidth memory capacity can’t scale to match the persistent context requirements of multi-turn inference workloads. We’ve hit the memory wall. Join Val Bercovivi, Chief Technology Officer of WEKA, as he breaks down how token warehousing can transform AI economics,and why organizations that architect memory persistence as an infrastructure priority, rather than an afterthought, will win out.

As AI investment accelerates, and as agentic AI enters the mainstream, business leaders are searching for the most effective way to seize this historic moment. Many have high expectations – and significant investments planned – for AI in 2026 and beyond. But here’s the hard truth: there’s no silver bullet, no “plug and play” solution, no shortcut to scalable AI success. So what does today’s enterprise need to get right to get AI right? To unlock AI’s true potential, a bold data strategy should become a top-line business priority, and should align with business value. Join Ashish Verma, U.S. Chief Data & Analytics Officer of Deloitte, as he breaks down the key enablers of enterprise AI success.

You've heard the experts, now join this corresponding dynamic networking session to continue the conversation and ensure meaningful connections that support your business goals:Meet the expert speakers to get your questions answeredConnect with peers to discuss the key talking pointsUncover the technology solutions to the challenges presented in the debate

Workforce challenges behind digital infrastructure have become more complex and urgent due to unprecedented industry growth. Workforce strategies remain fragmented as companies are required to reskill at scale while also focusing on future skills and evolving education programs. This debate examines who owns the talent pipeline, how non-traditional talent can be developed, and how AI is reshaping the current skill set requirements.How can veterans, career-changers, and tradespeople be integrated into the data center workforce?How can K-12, higher education, and industry better align to prepare the next generation of data center engineers and operators?Which roles will evolve or be redefined as AI adoption accelerates?

Managing Director of Digital & Compute Infrastructure Moonshot Energy As AI workloads continue to place strain on hyperscale data centers, the need to move workloads closer to networks, data sources, and end-users is becoming more critical by the day. Join us for this exclusive deep-dive into a partnership aimed at addressing these very latency, cost, and sovereignty constraints in a way that may be more capital efficient and compute efficient, as we make the shift from training to inference.

Explosive demand for digital services and distributed compute is exposing connectivity bottlenecks across fiber networks, interconnection points, and regional backbones in North America. As data centers, telcos, and cloud providers scale faster than networks can expand, the race is on to deliver low-latency, high-capacity infrastructure that reaches new regions without compromising performance or compliance. This debate brings together telecom and data center leaders to explore how collaboration, investment, and innovation can close the connectivity gap.In this discussion, they address:Where new workloads and applications are straining existing network capacityHow operators can collaborate across fiber, subsea, satellite, and wireless systemsThe role of regional hubs and cross-border connectivity in reducing latencyHow regulation, spectrum policy, and incentives can accelerate buildoutsWhich emerging technologies will define the next phase of U.S. connectivity

With AI-scale demand surging and government priorities shifting away from sustainability, data centers stand at a crossroads: take the quick path to power, or the responsible one - which comes with greater uncertainty. Gas delivers megawatts quickly, but at the cost of emissions and reputation. Renewables struggle with cost, intermittency and grid integration, whilst the promise of SMRs and hydrogen are still too far out to count. This debate will examine whether the U.S. can expand digital infrastructure at speed without deepening its carbon footprint.This debate will address:How shifting state and federal policies are affecting the U.S. energy landscapeThe trade-offs between reliability, reputation and emissions in power procurementWhat are the options for renewables that deliver a cleaner baseload?Is natural gas the only bridge fuel for US data centers?Will speed trump sustainability as we look to meet demand for AI?

As data centers grow larger, more connected and more visible, they are exposed to a greater and greater number of security threats - both physical and cyber. In many cases the line between different types of security is blurring, meaning true resilience must be designed into the fabric of a data center, rather than added as an afterthought. This debate explores how operators can align cybersecurity, physical protection, and governance into a unified strategy that safeguards uptime, reputation, and trust.This debate will explore:How to integrate cybersecurity and physical protection from concept to operationManaging third-party and supply chain vulnerabilitiesHow evolving regulation and public scrutiny are reshaping security prioritiesThe trade-offs between accessibility, transparency and defenseWhat a “security-by-design” model looks like for tomorrow’s data centers

Head of Datacenter & Energy Infrastructure Research SemiAnalysis As new cloud providers enter the market to address the seemingly insatiable demand for GPUs, there are many possible trajectories for these new players across a multitude of enterprise verticals to partnerships with hyperscalers and colocation providers. How will these new GPU-cloud providers evolve? What, ultimately, will make or break the success of these new business models?

You've heard the experts, now join this corresponding dynamic networking session to continue the conversation and ensure meaningful connections that support your business goals:Meet the expert speakers to get your questions answeredConnect with peers to discuss the key talking pointsUncover the technology solutions to the challenges presented in the debate

Connect with other industry leaders for over lunch

As AI workloads continue to dominate, massive amounts of data is required to move between different regions, cloud, or on-premise servers, which is slow, expensive, and manually intensive. How can this process be optimized without actually accessing the data itself? Join us for this special look at Sparta, as they share how they aim optimize data pipelines with a single-line-of-code integration without manual tuning, pipeline rewrites, all without ever decrypting or viewing the underlying content.

This taster session introduces the role of liquid and precision cooling in maintaining optimal thermal thresholds in data centers, enabling learners to identify the most appropriate technology for different environments whilst taking into account sustainability, efficiency, and operational cost.Designed for professionals across the data center ecosystem, this session supports learners in understanding the latest industry trends, including the growing adoption of liquid cooling in response to high-density and AI-driven workloads.The session includes a 45-minute overview, interactive Q&A, and networking opportunities, and is open to all attendees of DCD>Connect New York.

As AI demand accelerates, utilities are tightening terms while developers race to secure capacity, often ahead of real demand. The market is shifting from “pay as you use” to “pay whether you use it or not”, as hyperscalers reserve vast power blocks while interconnection queues grow and local communities question energy use, costs, and land impact. This debate brings together utilities, cloud providers, and financiers to ask whether capacity is being built in the right regions and how current strategies affect grid planning and investment in 2026 and beyond.This discussion will address:How AI workloads are changing regional power demand and siting decisionsWhat overbuilding and stranded capacity mean for long-term investmentHow regional differences shape interconnection timelines and project viabilityHow financing terms and take-or-pay models are influencing developer behaviourWhat smarter, data-driven capacity planning could look like in the AI era

Edge computing is no longer a speculative future - it’s becoming business as usual. As AI inference, real-time analytics and IoT workloads proliferate, companies are deploying compute resources closer to users and devices. Solutions now prioritize ultra-low latency, data sovereignty, and localized decision-making rather than centralised cloud only. This debate will bring together operators, regulators and enterprise leaders to assess which workloads truly benefit at the edge, how the business case is evolving and how infrastructure must adapt accordingly.This debate explores:Which workloads we need to de-centralize and whyHow regulation and data-sovereignty concerns shape edge vs cloud placementThe infrastructure model for edge deployments: size, location, connectivity and costHow security, resilience and manageability differ at distributed sites vs central cloudsThe USPs of edge for finance, healthcare, media and other industries

Rising compute and storage densities are reshaping how data centers are built from the rack up. The once-stable balance between power, cooling, and network design is giving way to a new era where efficiency, integration, and bandwidth are pushed to their limits. As performance demands surge, operators and architects are reimagining how servers, interconnects, and fabrics can be designed to deliver more capacity within the same physical footprint. This debate explores how far existing rack architectures can stretch - and what new approaches will define the next generation of data center design.This debate will explore:How high-density compute is driving changes in rack layout and connectivityThe trade-offs between bandwidth, latency, and efficiency in rack designHow to evolve power delivery, cooling, and cabling for next-gen hardwareWhether today’s rack designs can scale with the pace of modern workloads

Canada is emerging as a critical hub for AI and hyperscale computing, but as US-based operators expand north, questions around data sovereignty, compliance, and cross-border workloads are becoming increasingly complex.This debate explores who will ultimately shape Canada’s data center market, and how operators can navigate cross-border complexity, align compliance with strategy, and ensure Canada claims can claim a spot as a true global data center leader.This debate will address:How operators can ensure AI workloads comply with Canadian privacy and security laws while leveraging Canada’s energy and fiber infrastructure.The practical challenges of storing, processing, and transmitting AI data in CanadaHow cross-border collaboration with US data centers and networks affects market strategy and sovereigntyThe trade-offs between compliance, competitiveness, and building a long-term Canadian data center powerhouse

GPU clouds are now at the center of the AI boom, representing the most important transactions in AI. What started as a repurposing of GPUs used for crypto mining has transformed into a pillar of the AI ecosystem. By tapping into unmet demand for specialized workloads that require specialized hardware, neoclouds are set to transform the marketplace for compute. Join us as we sit down with true neocloud leaders as they weigh in on the future for enterprise buyers, from financial services to pharmaceuticals.

With the majority of US-based executives reporting adoption in their companies, and agents moving from mere conversational responses to active task execution, what will 2026 bring for the adoption of agentic AI across enterprise? Which verticals will see the most immediate impact? And will this be the year to overcome the "pilot to production chasm"? Join us for this exploration of key case studies for enterprise AI adoption.

Global Head of Data, AI/ML, GenAI & Advanced Computing AWS The next frontier of AI is not defined by where the cloud ends, but where the intelligence begins. As generative AI moves from massive data center training to real-world execution, the architecture of "intelligence" is being radically redistributed. Join Ramya Winstead, Global Head of Data, AI/ML, GenAI & Advanced Computing at AWS, as she breaks down how the shifts towards agentic AI and industrial automation are driving the convergence of cloud, edge, and in-device computing.

Is the closed AI stack starting to crack? Closed ecosystems come with hardware compatibility issues, costly licensing, and innovation bottlenecks, leading to spiralling infrastructure costs. The optimal future for AI computing is an era where open source meets open hardware. And more importantly, where customer-centric partnerships enhance performance of open compute. Join Darrick Horton, CEO of TensorWave, as he makes the case for an open, cooperative computing future.

The ability of enterprises to scale AI is constrained by both physical hardware investment and the need for compliance-ready infrastructure. How can regulated industries scale AI workloads dynamically? And what exactly are regulated industries doing with their cloud GPUs? Join Travis Cannell, COO of Vast.ai, as he breaks down the AI use cases of regulated enterprises, and why isolated infrastructure and full data control enables compliant scalability.

By 2026, the AI hardware landscape has shifted from a "buy-everything" sprint to a strategic marathon. As Agentic AI and Small Language Models (SLMs) prioritize real-time inference over massive training runs, the criteria for "the right chip" are changing. Join us to explore how the inference shift is diversifying procurement away from general purpose GPUs, and what you can expect from the future GPU supply chain.

You've heard the experts, now join this corresponding dynamic networking session to continue the conversation and ensure meaningful connections that support your business goals:Meet the expert speakers to get your questions answeredConnect with peers to discuss the key talking pointsUncover the technology solutions to the challenges presented in the debate

In the era of explosive AI growth, scaling infrastructure to meet the demands of massive models and real-time workloads requires a new paradigm: the AI factory. This session, led by CoreWeave's Director of Engineering, delves into how purpose-built cloud platforms are transforming data centers into efficient, GPU-optimized powerhouses capable of handling exascale computing. Learn about real-world deployments, such as the world's first cloud rollout of NVIDIA's GB300 NVL72, and strategies for building multi-gigawatt AI factories in partnership with NVIDIA to accelerate training, inference, and innovation.

For decades, the power of network telemetry has been locked behind domain expertise. Protocols like NetFlow, IPFIX, and sFlow are extraordinarily rich, capable of revealing traffic patterns, performance bottlenecks, cost drivers, security anomalies, and application behavior. But unlocking that power has traditionally required deep networking knowledge. You had to understand what the protocol could expose, how to query it, how to aggregate it, and how to interpret it. After more than a decade working directly with customers, I have seen this journey firsthand. Organizations invest in collecting high-fidelity flow data, but extracting value requires time, training, and iteration. Teams must learn how to ask the right questions, build the right views, and interpret the outputs correctly. Even then, network experts are repeatedly pulled into calls to answer simple questions because others in the organization lack the domain context to navigate the tools. The result is that one of the most powerful datasets in the enterprise often remains underutilized. This session explores the shift from dashboards to dialogue — where AI becomes the interface to network telemetry. Instead of requiring protocol fluency or query language expertise, AI allows users to express intent in natural language. It translates that intent into structured analysis, dynamically correlates data, and surfaces insights that were previously hidden behind layers of specialization.

AI-driven workloads are reshaping data center infrastructure. GPU clusters create rapid swings in power and thermal demand, yet many facilities still rely on polling-based monitoring designed for slower, static environments. This session examines the architecture required for continuous, real-time telemetry in AI-scale facilities. We focus on:

Each of our partners has 5 minutes and 5 slides to deliver their boldest ideas and game-changing cooling solutions. Fast-paced and impactful, this is the place where the future of digital infrastructure takes shape.Hear the experts pitch their latest innovations, then join our dedicated 'Explore the Floor' session - your chance to connect with the speakers and dive deeper into the pitches that grabbed your attention!

Each of our partners has 5 minutes and 5 slides to deliver their boldest ideas and game-changing design, build, and operations solutions. Fast-paced and impactful, this is the place where the future of digital infrastructure takes shape.Hear the experts pitch their latest innovations, then join our dedicated 'Explore the Floor' session - your chance to connect with the speakers and dive deeper into the pitches that grabbed your attention!

Each of our partners has 5 minutes and 5 slides to deliver their boldest ideas and game-changing power and sustainability solutions. Fast-paced and impactful, this is the place where the future of digital infrastructure takes shape.Hear the experts pitch their latest innovations, then join our dedicated 'Explore the Floor' session - your chance to connect with the speakers and dive deeper into the pitches that grabbed your attention!

What does it actually take to build an “AI factory”? How do you plan for an ever-shifting landscape of rapidly evolving AI models, compute capacity, and hardware? Join us for this special fireside chat as we explore what building the data centers of the future really means.

As AI workloads drive unprecedented investment in datacenter infrastructure, teams are being pushed to compress the design, build, deploy, operate loop at a pace and scale that defies traditional systems and processes for datacenter management. From hyperscalers to neoclouds, teams must bridge physical and logical designs to unlock effective automation and manage massive footprints with high rates of change. Aligning data across management systems is an enormous barrier to velocity for operators. Physical designs must align with component level plans, which must feed BOMs, which must align with shipping manifests, which must drive field operations and integrate with provisioning systems, which must couple into configuration automation tooling for devices and observability, and so on. This session explores how the world's best AI datacenter teams are adapting their operations to leverage consistent data, driven by intentional design, to ensure cohesion and unlock speed and leverage at massive scale. We'll deliver a practical look at what it really takes to manage accelerated compute environments in 2026.

The potential for AI to become a profit generator lies beyond mere productivity and efficiency gains. AI systems have the capability to transform the future of AI at work, especially in sales, where context, judgment, and institutional knowledge are critical. Join Ari Lehavi, Head of Applied AI at Moody’s, as he shares how AI is being used to turn top performers into force multipliers.

Unprecedented growth in AI forced the industry to focus on AI infrastructure for training to start with. We made tremendous progress, led mostly by hyperscalers, in optimizing AI training by balancing compute and communication in large uniform clusters. Now the focus is shifting to enterprise use cases where we have to fine tune and deploy these foundational models to serve our customers. The challenges are even bigger as enterprises deal with heterogenous infrastructure, lack of AI talent with deep understanding and operational challenges with power, cooling etc. This presentation delves into the insights from hyperscalers, neo clouds and enterprises in making intelligent choices to create business value.

AI application development teams are often hindered in terms of access to networking and storage performance, which can often pale in comparison to the powerful compute they have access to. Join WhiteFiber Chief Technology Officer, Tom Sanfilippo, as he details the importance of matching powerful compute with the right networking and storage capabilities for customers large and small.

Chief Information Officer U.S. National Nuclear Security Administration As AI revolutionizes both the everyday lives of the consumer and the day-to-day operations of the enterprise, how can U.S. governments enhance the capabilities of government workers to make a bigger impact serving the American people? And what considerations exist for government bodies when it comes to laying the hardware foundation for AI deployment?

Distinguished Engineer - AI/ML & Data Analytics Platform Engineering/Architecture TD The potential impact of AI on enterprises is incalculable - from productivity and efficiency gains from automation to AI-driven customer experiences to drive customer satisfaction and retention, the potential benefits are too many to count. So how can you turn potential benefits into actual ones? And how can you measure the real value extracted from AI within today’s enterprise? And how should you think about your infrastructure, from cloud to on-prem, in order to optimise for AI and traditional workloads? Join us for this fireside chat as we explore exactly how AI success can be benchmarked and enjoyed.

Compute is arguably today's hottest commodity, and yet it's not yet treated as such. And given that compute powers the AI infrastructure boom, is it time for the introduction of a new market structure to manage risk and bring liquidity and price discovery to this new landscape? How will this new financial infrastructure behind digital infrastructure look? What does this mean for investors, data center operators, and regulators? And will storage become the next hot commodity? Join us for this exclusive panel with the minds behind this innovative new approach to digital infrastructure finance.

The biggest bottleneck is AI today isn’t GPU availability. It’s the pipeline feeding them. If your storage architecture isn’t leverage parallel systems and RDMA, you’re essentially putting bicycle tires on a Ferrari. Join us for this in-depth look at the role of storage as the heart of the AI factory stack, and why it needs to be as big a priority as your compute.

AI data centers are scaling to gigawatts. Capex is measured in tens of billions. Memory bandwidth is the bottleneck. Power is the constraint. Now quantum wants in. But here’s the uncomfortable question: does quantum integrate into AI factories as a new accelerator class , or does it remain a sovereign science experiment parked next to a national lab? Logical qubits don’t run alone. They drag along massive classical overhead: control electronics, GPUs for error correction, memory footprints that scale nonlinearly, cryogenics that don’t fit neatly into standard rack designs. The physics is hard. The infrastructure might be harder. This session examines the real intersection: hybrid orchestration, classical amplification effects, facility integration, and whether AI data centers become quantum co-location hubs. No qubit-count vanity metrics , just system architecture, power math, and deployment reality.

As nations looks to build out their own sovereign AI infrastructure, Canada is surging forward with a $705 million commitment towards an AI Sovereign Compute Infrastructure Program (SCIP), with Canadian companies using Canadian-built AI systems and sovereign supply chains to suit critical data and compute that remain governed in Canada. What are the strategic aims of such projects? How will they seek to address the risks associated with an over-reliance on non-Canadian infrastructure? And what does the journey look like for companies moving forward as part of this grand project - from investment to data centres to filling the data centres with GPUs?

How can enterprises and service providers move beyond experimentation and into standardized, scalable AI infrastructure models? Join Wellington Lordelo, Global Director of AI & Innovation at Digital Realty, as he offers a practical blueprint, exploring AI Factories vs Gigafactories, validated partner ecosystems, and more.

Operators are afraid their data centers won’t get built. Why? Community backlash. Local residents and businesses worry that data centers will consume too much electricity and drive up utility prices, but create fewer jobs and economic benefits than other industries. There’s a solution: pair data centers with advanced nuclear power. Advanced nuclear projects bring what every community wants: good jobs, new tax revenue, and new reliable generation that eliminates competition for power. Oklo’s landmark deal with Meta in Pike County, Ohio is a perfect example. The multi-year project will create thousands of long-term jobs, pump tens of millions of dollars into the local economy, and deliver safe, clean, 24/7 power. The economic benefits of advanced nuclear power are changing people’s minds: from backlash to buy-in, from “Not in my backyard” to “Build it here!”What you’ll learn:How advanced nuclear reshapes the local value proposition for data centersWhy States and local communities are competing to attract advanced nuclear projectsWhat the Oklo-Meta project in Ohio teaches us about winning local support for data centers

Following our 5x5 Lightning Pitches, it's time to 'Explore the Floor.'You’ve just heard the Lightning Pitches, and now is the time to get your questions answered and dive deeper into the latest tech innovations at our partner stands.

Member of Technical Program Staff - Infrastructure Strategy & Capacity Planning OpenAI Building the infrastructure to support the next generation of frontier models requires more than just silicon; it requires a fundamental reimagining of the global energy grid and physical supply chains. Join us for this deep-dive into what it takes to procure for the largest digital infrastructure projects ever created.

Capacity constraints have dominated headlines for years, and with demand continuing to soar, operators are being forced to redefine what resilience really means, and adapt their strategies accordingly. In the face of power access challenges, cost volatility and the growing talent gap, how can operators develop the right mix of infrastructure, strategy, innovation and investment to build data centers that can endure as well as expand?This panel assesses the core requirements essential for long term resilience, including:Power: Securing affordable, reliable megawatts and strengthen grid partnershipsPricing: Managing volatile costs in materials, energy and essential technologiesPlaces: Rethinking siting land against the backdrop of tougher permitting and power scarcityProcess: Ensuring operation consistency and resilience across expanding portfoliosPeople: Building skilled teams and community trust to sustain long-term capacity growth

Principal Product Manager, AI Storage & Data Services CoreWeave A dedicated, intimate, peer-to-peer set of roundtable discussions for specialists in Storage (cloud or on-prem) and data management.

A dedicated, intimate, peer-to-peer set of roundtables for leaders from across the Quantum computing ecosystem.

A dedicated, intimate, peer-to-peer discussion for specialists in network hardware and software - for Network Engineering, Infrastructure, Operations, Architecture, and Automation.

Principal Product Manager, AI Storage & Data Services CoreWeave A roundtable exclusively for AI cloud/GPU cloud/neocloud ecosystem leaders and their industry partners, discussing the most urgent challenges and exciting opportunities.

Join your peers a for our close of day one Icebreaker Party!Connect with other industry leaders for an unforgettable evening of networking and entertainment - open to all attendees.

Kickstart your day with a 5K networking run or an inspiring Walk & Talk.Start the day right and join your peers for either a 5k wake-up run around Central Park, or for those who enjoy a slightly slower pace, a 2.5k Walk & Talk - meet at the monument in Central Park at 7am!If you have any questions at all please contact the team at [email protected]

Join us for a welcoming breakfast with the Armed Forces data center community at DCD New York. Hosted by FlexGen and iMasons Armed Forces, this event is open to all members of the Armed Forces community and their supporters. Connect with service members, veterans, and industry peers to network, share ideas, and foster collaboration across the data center ecosystem.

Delegates collect their badges and get their personal agenda and app questions answered.

The Show Floor opens and all partner Stands and Lounges are open for business!

The real demand for AI infrastructure is no longer coming from traditional hyperscalers, it’s coming from frontier AI labs. As companies like Anthropic and OpenAI push toward massive training runs and global-scale inference, they are increasingly dictating how data centers are designed, where they’re built, and how compute is deployed. This keynote explores how labs are reshaping infrastructure from site selection and latency-sensitive inference to ultra-high density data centers optimized for token economics, power delivery, and scale.

Chilled water distribution systems can be complex and difficult to manage.Join Grundfos to explore an alternative means of distributing chilled water for hybrid cooling in the data centre environment, delivering energy savings, overcoming balancing issues, and giving greater control over differential temperature.Distributed Pumping is a paradigm shift towards decentralized pumping.

Power has become the defining constraint for data center growth. AI and high-density workloads are forcing operators to move beyond traditional utility models, driving investment in private generation, private wire partnerships, hybrid PPAs and nuanced regional approaches. The challenge is no longer about just accessing power, it’s about how we can create, control and future proof it to make energy a competitive advantage, not a constraint.This panel explores how operators are navigating evolving power markets by addressing:The evolution of power procurement models to meet high-density demandNew connections: The rise of microgrids, private wire and direct renewable integrationThe trade-offs between deploying in constrained hubs versus new marketsThe new collaborations between utilities, developers and policy makers required to sustain AI growth

AI data center demand is booming. However, while GPU clusters can now be sourced in months, the lead time for the physical infrastructure required to both power and house them is growing exponentially and becoming the industry’s biggest bottleneck.Much of the equipment and components require skilled labor to produce, particularly welders, but according to the American Welding Society, the U.S. is projected to face a shortfall of 330,000-400,000 welders by 2028.With timelines slipping and construction costs rising, how do we scale physical construction to meet the demand?In this session, Path Robotics CRO Heather Caroll will share insights from current customers deploying Physical AI for manufacturing, specifically welding, and how it is offsetting labor shortages while increasing throughput of critical data center and grid components, enabling industry leaders to build at the speed the AI economy demands.

Chief Technology Officer, Livermore Computing Lawrence Livermore National Laboratory

Cooling is no longer just an operational necessity, it is a strategic lever.This session presents Cooling-as-a-Service (CaaS): a pragmatic, outcome-focused operational model. Beyond operations and service, CaaS encompasses a full range of solutions spanning the entire data center, including CDUs, coolant health monitoring, coolants, analytical services, and more. CaaS shifts responsibility for fluid health monitoring and lifecycle care to a specialized provider, while operators maintain control over SLAs and integration. Explore where CaaS fits in the modern data center ecosystem, how it compares to traditional approaches, and key considerations when assessing service-based cooling strategies.

As AI workloads push data centers into unprecedented levels of density, complexity, and power demand, traditional infrastructure approaches are no longer enough.Join Michael Giannou from Honeywell to explore how integrated building automation forms the backbone of a resilient, scalable, AI‑ready data center.Attendees will learn how combining electrical systems with building automation can enable real‑time visibility and predictive maintenance capabilities that are essential for maintaining uptime and driving operational efficiency.

The rapid innovation cycles for AI data centers and co-design specialization, while boosting performance, can strain multivendor supply chains, leading to higher costs and reduced efficiency. The industry must tackle this by fostering commonalities and an evolvable strategy for AI infrastructure within the OCP Community, mirroring its past success with cloud data centers. A number of hyperscalers, neoclouds, and AI leaders have publicly supported this effort by authoring and signing an OCP Community open letter. This letter calls for increased open collaboration, focusing on greater fungibility for data center physical facilities and grid interconnects, as well as more open IT hardware and systems management specifications for AI. The OCP Foundation will present an overview of how open source collaboration has historically succeeded in data centers and discuss its application to the emerging opportunities and challenges of AI data centers. The presentation will analyze the infrastructure requirements, identifying areas for near-term open source and standards adoption—the "low-hanging fruit"—and those requiring more time for maturity and convergence. This analysis will lead to a comprehensive enumeration of the many OCP Community alliances, projects, and initiatives that are actively shaping the future of open and evolvable AI infrastructure. These efforts span critical areas, including higher power distribution, liquid cooling, servers, switches, storage, silicon, and the systems management for both IT and operational technology (OT) components, concluding with ways the audience can get involved.

The greenest data center isn’t necessarily the one with the lowest emissions, but the one that gets the most compute from every watt, and the longest lifetime from any component. Sustainability is shifting towards smarter and more efficient practices across energy management, water usage, grid, onsite resources and beyond. Operators are rethinking how power, cooling, and compute interact. This panel addresses how efficiency is reshaping sustainability strategies, and where innovation matters most in the next phase of data center design.Which technologies are offering the greatest efficiency gains?How operators are reducing carbon emissions and water useAre KPIs metrics aligned with efficiency strategies?How resource and cost efficiency go hand-in-hand

Even with sites, capital, and demand lined up, many data center and enterprise projects still stall on one thing: power.Join ECM President Eugene Garcia to explore their Power Advantage Flywheel, an ISO-direct framework that helps large energy users access capacity sooner by knowing how to navigate the interconnection process, reduce delivered energy costs by up to 25%, align carbon goals with financially sound wholesale strategies, and create new energy related revenue streams that were previously unattainable.You’ll leave with a clear framework for where your organization enters that cycle and which ISO-direct levers can turn power from “the constraint” into a repeatable advantage.

Future builders. Future operators. Future leaders.This dedicated workshop provides a practical pathway into the data center ecosystem for the student community. Through industry 101 sessions and firsthand shared career journeys and experiences, this interactive day will demystify the digital infrastructure landscape on which the world relies, and outline the career opportunities available to emerging talent.Powering the Next Generation – Student Workshop Agenda10:00 AM – 10:20 AM: Welcome & Check In10:20 AM – 10:40 AM: Opening RemarksCyre Mercedes Quinones, CEO, iMasonsHiba Agha, Director of People Programs, iMasons10:40 AM – 11:00 AM: Introduction to the Nomad Futurist AcademyNabeel Mahmood, CEO & Co-Founder, Nomad Futurist Phillip Koblence, COO & Co-Founder, Nomad FuturistKelly Santalucia, Executive Director, Nomad Futurist 11:00 AM – 11:30 AM: Inside Digital Infrastructure: The Backbone of Our Connected WorldPhill Lawson-Shanks, Chief Innovation Officer, Aligned Data CentersIlissa Miller, Founder & CEO, iMiller Public Relations (iMPR)Vic Rose, Director, AI Inferencing, MicrosoftAdam Noll, Founder, Edgeology11:30 AM – 12:00 PM: Q&A & Open Discussion12:00 PM – 12:30 PM: Scaling Clean Power, Scaling People: Creating a Sustainable Workforce for the Digital Infrastructure EraHiba Agha, Director of People Programs, iMasonsJoel Chakkalakal, Senior Solution Architect, Critical Risk SolutionCy Sinclair, Data Center Manager, Bloomberg12:30 PM – 1:00 PM: Networking Lunch1:00 PM – 1:30 PM: Power, Water & People: How Digital Infrastructure Can Grow ResponsiblyAshley Dirou, Senior Sales Developer, Data Centers, GrundfosMichael Obradovitch, Vice President Global Accounts - Data Centers, EcolabMelissa Farney, Director of Marketing, TECfusions1:30 PM – 2:00 PM: From Curiosity to Career: Your Path Into Digital InfrastructureAlison Marlow Reese, Data Center Business Development Leader, HDRMarc Bhuyan, Product Manager, GoogleIndigo Pinto, Emerging Markets Program Director, Bureau Veritas Primary Integration (BVPI)Harris Hamid, Student, Stevens Institute of Technology2:00 PM – 2:20 PM: Closing Remarks2:20PM – 3:00 PM: Q&A & Open DiscussionOpen to all students.

AI has accelerated demand for high-density infrastructure, but many enterprises don’t have the luxury of greenfield construction—or the desire to overinvest before proving ROI. In this session, we’ll discuss how organizations can leverage and transform their existing data center assets to support AI initiatives. Attendees will learn how to evaluate retrofit investments, determine when software‑driven optimization can defer capital spending, and build an ROI model that reflects both physical constraints and strategic opportunities.We’ll highlight the emerging importance of digital twins and advanced modeling tools that help leaders make informed decisions about power usage, electrical paths, cooling resilience, and risk. Whether you’re planning your first GPU cluster or scaling AI workloads across your footprint, this session will help you align infrastructure strategy with business outcomes.

The artificial intelligence industry is undergoing a critical transition, shifting focus from model training to massive-scale inference. As AI capabilities move towards sophisticated, real-time reasoning and autonomous systems, the demand for specialized, low-latency, and cost-effective inference compute has exploded, creating new opportunities for new players in the market for compute. Join us for this unparalleled deep dive that moves past the "training wars" to explore who is best positioned to power this new inference era and how "co-opetition" is reshaping partnerships, software ecosystems, and hardware roadmaps in the race to build out compute for every AI factory.

The AI infrastructure buildout is the largest, fastest expansion the data center industry has ever seen. Major cloud providers are expected to spend over $600 billion in capital expenditures in 2026, with roughly $450 billion directed at AI infrastructure alone. That scale of investment has created cascading pressure throughout the IT hardware supply chain.Previous-generation hardware that would have been written down to near-zero is finding eager buyers across a growing ecosystem of AI startups, research institutions, and mid-market enterprises racing to deploy their own AI capabilities.In this session, James Patrignelli from Liquid Technology, a leader in IT Asset Disposition with over 25 years of experience turning excess IT equipment into working capital, breaks down what’s happening in the supply chain, who’s buying this hardware and why, and how data center operators can rethink their refresh and disposition strategies to capitalize on a market that may not look like this forever.

Onsite power generation solutions using natural gas and other fuels have become a trusted alternative for large load data centers serving artificial intelligence (AI) and other applications. The IT compute, cooling, and other power needs of a single site can require 100-1000+ MWs of capacity and many 9’s of reliability. AI load profiles can be volatile, fluctuating wildly in milliseconds or faster. Because of this, traditional power generation is challenged to respond fast enough to maintain IT-required power quality without tradeoffs such as operation outside of the equipment “sweet spot” which may cause sub-optimal performance, emissions, fuel consumption, maintenance, and other lifecycle cost.Wärtsilä’s hybrid power approach uses medium speed reciprocating internal combustion engines (RICE) together with Wärtsilä Energy Storage Systems and award-winning energy management software (GEMS) to address this challenge. The engines provide dependable and steady power, while the battery energy storage systems (BESS) dispatch power to “smooth” power fluctuations. This combination creates a stable power flow, reduces stress on equipment, and supports operation that meets environmental requirements even during the most demanding AI training cycles.Wärtsilä’s deployed capacity of 79 GW of engines and 20 GWh of energy storage enable flexible, reliable hybrid power systems that stabilize rising AI datacenter loads, ensuring continuous, high availability operation as demand and volatility accelerate.

AI, electrification and re-industrialisation are creating a new phase of data centre growth on both sides of the Atlantic. At the macro level, the narrative is clear: exponential compute demand, hyperscale expansion, capital abundance. But data centres do not connect to “global grids” — they connect to local substations, local regulators, and local politics. In the US, state-level energy regulation, transmission constraints and community pushback are reshaping siting decisions. In Europe, net-zero commitments, market design reform and power price volatility are colliding with industrial policy and digital sovereignty ambitions. This session explores the growing divergence between the logic of global capital allocation logic and local regulatory reality. The winners of the next cycle will be those who can translate global strategy into local execution.

As compute demand accelerates, driven by AI, analytics, and data-heavy workloads, enterprises are being forced to rethink what their cloud strategies were built for. Traditional architectures optimized for scalability and cost are no longer enough - flexibility, performance, and workload placement now define competitiveness. Current business models are forcing a next generation of compute that requires a balance between different environments, along with smarter approaches to data locality, cost control, and energy efficiency. This panel brings together industry leaders to explore how hybrid strategies are evolving to meet the performance, and governance needs of AI-era workloads.We will address:How AI is reshaping cloud architecture and investment decisionsBalancing performance, cost and sovereignty across hybrid and multi-cloud environmentsWhen to deploy workloads across cloud, on-prem and edge for optimal efficiencyManaging brownfield footprint while cloud strategy is in fluxWhat defines a truly “next-generation” cloud strategy in 2026 and beyond

The digital infrastructure industry is feeling real pressure: demand is rising, supply chains remain unpredictable, and expectations around speed and visibility keep increasing. The bottlenecks across procurement, manufacturing, and delivery are harder to ignore.For years, orders have moved through opaque systems, manufacturing capacity has been fragmented, and supply strategies have stayed reactive. That approach no longer works. Compressed timelines and higher performance expectations require a different operating model.This session focuses on what needs to be changed. Instead of managing bottlenecks, we’ll look at what replaces them: ecosystem-driven models that connect distributed manufacturing capacity with real-time data and full lifecycle visibility. When transparency and predictive insight are built into the operating model, infrastructure delivery becomes more coordinated, more accountable, and better equipped to scale.

As AI and high-density compute drive growth in data center power demand, utilities and operators are becoming increasingly tightly linked. However, they still plan and speak in different languages. Operators focus on workloads, SLAs, and speed to market. Utilities focus on grid stability, load profiles and regulated investment.This panel looks at how to build a common framework for fast changing data center loads. We’ll translate workload demand into grid terms, unpack queues, costs, timelines, and how earlier engagement with energy partners can de-risk projects and accelerate access to power.What utilities need beyond headline megawatts, such as load shape, flexibility, ramp rates, and growth profileOptions for operators to de-risk interconnection timelines and balance temporary generation with long-term grid upgradesCommercial models that align with incentives without weakening SLA commitmentsHow can collaboration be sustained across planning, construction and operations, and how can both sides define success?

As data centers scale in density and computing power, the need for resilient, predictable, and efficient thermal management has never been greater.Join Danielle Rossi from Trane to explore how chillers remain a foundational component of modern and future data center cooling ecosystems, especially within hybrid configurations that integrate liquid cooling, free cooling, and emerging heat‑recovery strategies.Join us to learn practical approaches, real‑world lessons, and forward‑looking perspectives on building robust thermal management architectures for the next generation of data centers.

As organizations and individual users continue to dramatically increase their use of AI, HPC, and even Quantum computing for applications ranging from automotive and aerospace to retail to making AI generated images of cats driving a sherman tank, how can our current digital infrastructure hold up? And how does the current trajectory of digital infrastructure evolution set us up for succeeding with these varied workloads? Join us as we explore the current and future bottlenecks, solutions, and the perspectives needed to transform the full stack.

The explosive growth of AI and cloud computing has pushed data centers to the forefront of the world’s energy and sustainability challenges.Understand how FuelCell Energy’s carbonate fuel cell systems can redefine what’s possible for modern data centers by delivering faster time‑to‑power, bypassing years‑long grid delays with rapid, on‑site deployment and modular N+1 redundancy, ensuring resilience as AI‑driven loads grow.This session will dive into our 1.25 and 2.5 MW designs, in addition to the new 12.5 MW FuelCell Energy Block that not only meets the needs of next-generation data centers, but enables them, offering continuous, scalable, near-zero emissions power that's built for the future AI era.

As data centers face unprecedented power demand driven by AI expansion, operators are increasingly dependent on highly reliable, digitally optimized electrical infrastructure. This shift places new emphasis on lifecycle service models that combine advanced diagnostics, predictive maintenance, and real‑time monitoring to ensure continuous, high‑uptime performance across substations and critical power assets.As grid interconnection timelines lengthen and on‑site power infrastructure grows more complex, data center operators must adopt service strategies that improve reliability, accelerate response times, and enhance asset longevity. Digital service solutions - integrating AI, analytics, and remote monitoring - play a central role in reducing unplanned outages, mitigating equipment risk, and enabling a more proactive and resilient operational environment. This session will explore how modern service frameworks are helping data centers scale more confidently, optimize asset performance, and prepare for the next generation of electrification‑driven growth.

As organizations and individual users continue to dramatically increase their use of AI, HPC, and even Quantum computing for applications ranging from automotive and aerospace to retail to making AI generated images of cats driving a sherman tank, how can our current digital infrastructure hold up? And how does the current trajectory of digital infrastructure evolution set us up for succeeding with these varied workloads? Join us as we explore the current and future bottlenecks, solutions, and the perspectives needed to transform the full stack.

Connect with other industry leaders for over lunch

Expect to mingle with the top-decision markers within the power ecosystem. From c-suite, policy makers, hyperscalers, energy suppliers, developers, utilities, investors, and data center operators over lunch. You’ll have actionable insights from all things power, ranging from: capacity planning, navigating hybrid energy offtakes, financing challenges, permitting struggles across the US, and best practices in navigating utility partnerships to help you navigate the race to power.

The Open Compute Project (OCP) is revolutionizing the way the industry thinks about compute infrastructure technology worldwide. OCP’s community-driven innovations use open standards to build a scalable, efficient and sustainable ecosystem where industry players collaborate within a safe framework, shaping the future of compute infrastructure. With the ever growing demands impacting the industry, there’s no better time than now to collaborate with the OCP Community. Interested in learning more about OCP and how you can get involved with this growing global community to learn, collaborate and contribute with other industry professionals? Join Michael Schill, OCP Senior Director of Community and Alicia Schap, OCP Community Operations Specialist as they discuss what's happening within the OCP Community to learn more and get involved.

AI workloads are accelerating data center buildouts globally, but every part of the network is hitting its breaking point.In this session, we’ll focus on two strained fiber segments in the modern data center ecosystem: ultra‑dense inside‑DC structured cabling, and performance‑critical metro/long‑haul DCI, now incorporating emerging technologies such as hollow core fiber for lower‑latency transport.Explore how automation, guided workflows, and continuous fiber monitoring allow operators to expand capacity rapidly without compromising quality.

Transmission congestion is extending time to power in the data center ecosystem, but Grid‑Enhancing Technologies (GETs) offer fast, cost‑effective relief without waiting for new transmission.By combining advanced conductors, dynamic line ratings, and efficient coatings, utilities can unlock substantial hidden capacity through improved thermal performance and real‑time visibility.Join Charles Holcombe from Southwire to explore how these tools, alongside complementary GETs, can rapidly expand usable grid capacity and accelerate the transition to a more flexible, decarbonized system.

AI work loads, grid congestion, long interconnection queues, and constrained markets have triggered the need for new methods of accessing power. This Oxford-style debate asks to address: New data centers should be built and operated behind the meter (BTM) instead of in front of the meter (FTM).Do sustainability optics push, delay, or block builds? Does BTM vs FTM improve or worsen it?What are the potential community benefits (job creation, heat reuse, reliability of services)?How are larger build data centers approaching BTM solutions and how can hybrid models best succeed?We’ll consider interconnection timelines, permitting, and the role of batteries in bridging constraints. Using Sli.Do, audience members will be able to pose their questions to the panel.

Cooling has become one of the most complex challenges in modern data center design. Water scarcity, new disclosure regulations and tougher environmental standards are reshaping how and where operators can expand, meanwhile, rising rack densities and AI-driven workloads are accelerating the shift toward hybrid and liquid cooling systems. As industry bodies such as ASHRAE update their liquid cooling guidelines and hyperscalers commit to zero-water goals, the industry must redefine what best practice means for thermal design, and balance scalability, sustainability and performance.This session will address:The regulations and disclosure requirements shaping cooling strategiesWhich cooling designs are gaining traction in water stressed regionsHow we can drive efficiency in hybrid coolingThe sustainability credentials of different coolant liquidsThe liquid cooling technologies driving efficiencies at scale

Gigawatt-scale AI campuses are being built at record speed. Most rely entirely on pipeline contracts for fuel reliability, but in a constrained grid, capacity on paper is not deliverability in practice. During Winter Storm Elliott, 16 Bcf/d of production vanished in 48 hours. FERC confirmed that 24% of all generation outages were caused by fuel supply limitations. The engines didn't fail, the fuel simply wasn't there. As hyperscale campuses grow beyond 300MW and into the gigawatt range, fuel assurance must evolve from a contractual assumption into a physical, testable capability.This session explores the role of on-site LNG storage as a multi-day reliability layer that strengthens gas-powered campuses, supports grid stability, and enhances long-term operational certainty.

As the data center landscape rapidly evolves, learning how to leverage renewable energy storage, both residential and commercial, is key.Join CBS Battery’s Eric Hill, who spent years in renewable energy before entering the data center industry, as he explores the parallels between these two sectors and shares key lessons from his experience in energy storage.Learn how to address volatility. Once driven by fluctuating fuel prices, policy uncertainty, and grid dynamics, it is now resurfacing in the form of unpredictable AI compute loads, tightening power markets, and heightened energy risk, requiring the same strategic forecasting and hedging disciplines homed in renewables.

Principal Product Manager, AI Storage & Data Services CoreWeave As AI workloads continue to grow in volume, the storage horsepower required to keep up also needs to grow. And the differing requirements for HPC versus AI training, versus AI inference, are creating variations of storage hardware and software to meet these varied, specialized needs. How must we think about data storage moving forward to ensure performance is maximised for specific workloads, and that the development of hardware and software does not focus too narrowly on one workload type at the expense of so many more?

As AI workloads continue to stretch networks to their very limits, and the shift to inference continues, how can network infrastructure and architecture keep pace? How does the transition from high-throughput, centralized batch processing to low-latency distributed real-time application delivery take place? How are data center networking fabrics being redesigned for AI? And how do we need to think about security when it comes to continuous data flowing across the network? Join us for our first networking panel as we explore the biggest implications AI has for network engineers, architects and solution providers.

From finance to pharmaceuticals, the importance of protecting data is paramount. Ensuring data is stored and used safely and securely, whilst using the data effectively for AI training and inference, is a difficult balancing act. How can enterprises balance the considerations for data governance with the drive to use AI across the business? And to what extent can AI be trusted with customer data right now?

You've heard the experts, now join this corresponding dynamic networking session to continue the conversation and ensure meaningful connections that support your business goals:Meet the expert speakers to get your questions answeredConnect with peers to discuss the key talking pointsUncover the technology solutions to the challenges presented in the debate

Data center energy challenges are evolving faster than traditional power architectures can keep up and are pushing operators to rethink how power is generated, stored, and controlled on campus.Join Zhuoning Liu from FlexGen to explore how microgrid-based, utility-scale design principles, combining campus-level batteries, grid-forming power conversion systems, and advanced microgrid controls, can unlock faster deployment, improve power quality, and enable flexible interconnection without sacrificing reliability.The result is a future-proof, asset-agnostic energy platform that delivers the predictability of vertical solutions with the flexibility and transparency hyperscale data centers now require.

When surging power demand outpaces the grid, innovation becomes mission critical.In this session, Langley will present a proven, behind-the-meter power solution providing scalable, stabilized, utility-grade power, exactly where and when it’s needed.Langley’s Power Solutions Division brings together three wholly owned subsidiaries, each a global leader in their field: Bergen Engines (formerly Rolls-Royce Bergen Engines) of Norway; Italian generator specialist Marelli Motori; and Germany’s gold standard in power conditioning and stabilization, Piller Power Systems.Designed for speed and scale, our modular solutions can be rapidly deployed up to 500+ MW and beyond, dramatically cutting lead times. While traditional grid connections can take a decade or more, Langley’s unrivalled time-to-power enables critical infrastructure to move forward, without waiting for the grid to catch up.

Record levels of capital are flowing into the North American digital infrastructure market, driven by AI-scale demand and the race to secure capacity. Infrastructure funds, private equity, and sovereign wealth investors are reshaping how projects are financed through long-term power agreements, platform investments, and strategic partnerships. Yet, behind the momentum lies a more complex picture of risk: constrained grids, shifting regulation, rising costs, and the growing threat of stranded capacity. This panel brings together investors, developers, and financiers to examine how capital is being deployed, structured, and safeguarded in 2026’s fast-changing market.In particular they address:The financing models driving AI-ready data center expansionHow regulation, tariffs, and ESG targets are influencing investment decisionsWhere overbuild and stranded capacity risks are emergingHow investors are adapting to power, permitting, and delivery challengesWhat is the runway for future data center investments?

SNIA returns to share how its Compute, Memory and Storage community is supporting the rapid rise of AI-based architectures. Scott Shadley will highlight key SNIA initiatives, including computational storage, memory and storage models, and interface taxonomy work, along with the market trends shaping modern Storage integration. This session provides a concise update on how SNIA’s technical efforts, including StorageAI align with the datacenter roadmap and the next generation of AI systems.

As AI data center networks face demands to scale in multiple dimensions, from front-end to back-end, and from scale-out to scale-up and scale-across, Ethernet is poised to take the lead across all of them. Ethernet has emerged as the scalable, open, and cost‑effective foundation for modern AI networking, adapting to the unique demands of AI networking while leveraging a massive, proven ecosystem. This session explores how Ethernet is increasingly powering the next generation of AI data centers and what that means for real‑world deployments.

Permitting is a crucial component of US data center growth, ranging from site control to energization. This expert-led soapbox will showcase how interconnection queues, state siting rules, and local politics can translate into schedules and cost risk. We’ll compare East Coast, Texas, and West Coast markets.Differences in regional permitting models between PJM, NYISO, ERCOT, and CAISOHow do interconnection milestones impact investment timing?Are interconnection milestones pushing energization or is the timing risk shifting towards developers?We ask our experts: if you could change one policy to cut the data center schedule, what would it be and why?

The sheer volume of data collected by software systems such as DCIM and Digital Twins can quickly become overwhelming. That is why traditionally they are used to monitor and alert only.Join Nicholas Barrowclough, President at AKCP, to learn about how through the use of a network of comprehensive sensors you can leverage the power of “Physical AI”.Facts are a harness against opinions. Restrain the opinions with facts to enhance data center efficiency.

Rising power demand and grid interconnection delays are creating significant challenges for data center development across the United States. Increasing power density, projected to reach up to 600 kW per rack within the next two years, combined with lengthy ISO interconnection timelines, is pushing project approvals out by years.As a result, on-site power generation is becoming a critical solution. By 2030, an estimated 38% of new data center developments may incorporate on-site generation, with up to 27% potentially operating fully off grid.Flexible, rapidly deployable on-site power solutions are emerging as essential infrastructure to bridge power gaps, enable growth, and ensure reliable energy availability when and where it is needed most.

Billions are flowing into superconducting, photonic, neutral atom, and trapped ion systems. Every roadmap claims scale. Every investor deck implies commercial inflection is near. But architectures are not equal : and timelines are not honest. The real question isn’t just which modality reaches logical qubits first. It’s which one can scale economically, integrate with classical infrastructure, and deliver repeatable commercial advantage. Superconducting leverages semiconductor fabrication but carries heavy cryogenic and control overhead. Photonic promises networking elegance but faces integration complexity. Neutral atom and ion systems tout coherence and fidelity, but what does scaling actually look like at data center scale? And how much classical compute does each require? This panel forces two debates into one: Which modality has the cleanest path to practical utility , and is that utility 2-5 years away, or 15-20? What milestone would constitute “real” advantage? Where does classical AI acceleration erase quantum’s edge? Expect architecture-level analysis, power math, and disagreement, not roadmap optimism.

As AI models scale in size and complexity, traditional server architectures struggle to keep pace with growing memory capacity, bandwidth, and utilization demands. Compute Express Link® (CXL®) introduces memory expansion capabilities that enable more efficient, flexible, and scalable infrastructure for AI training and inference. This session will provide a high-level overview of CXL technology and how it helps accelerate AI workloads by reducing memory bottlenecks, improving resource utilization, and enabling composable, disaggregated systems. Attendees will learn how CXL-based memory sharing and pooling can reduce costs, improve performance, and support larger models across heterogeneous compute environments.

Director, Architecture & Strategy & Chairperson AMD & UALink Consortium AI infrastructure is entering a new phase where performance is no longer limited by compute alone—it’s limited by how efficiently accelerators communicate. As models scale from billions to trillions of parameters and clusters expand from racks to campus-scale deployments, interconnect architecture has become one of the most strategic design decisions in modern computing. Kurtis Bowman, Chair of the UALink Consortium, will introduce UALink and explain how an open, high-performance scale-up fabric is reshaping the future of AI systems. Attendees will gain a clear understanding of where UALink fits within the interconnect landscape, how it complements existing standards like PCIe, CXL, and Ethernet, and why open ecosystems historically outperform proprietary approaches at scale. The session will provide architects, system designers, and infrastructure leaders with actionable insights into emerging accelerator connectivity models, multi-vendor system design strategies, and the roadmap toward interoperable AI infrastructure. Investors and industry strategists will also gain perspective on how open interconnect standards can expand market opportunity, reduce supply chain risk, and accelerate innovation across the ecosystem.

President & Managing Director, EMEA; SVP Technology & AI Supermicro There have been, and still are, many bottlenecks when it comes to AI. First it was the obvious one, GPUs. Power remains a constraint, and so too does land. But perhaps the bottleneck that only a few saw coming but now everyone sees, is storage and memory. Hard drives are all sold out for 2026. RAM prices continue to rise as storage persist, as demands have skyrocketed and there is no end in sight. Join us for this expert-led panel as we break down what led us here, and what we do now.

Join us for a networking coffee to continue the conversations sparked during the Energy Summit, offering a space to connect, reflect on key insights, and explore collaboration opportunities. Aligned with WRISE’s mission to advance an equitable energy future, this gathering encourages relationship building and shared learning across the renewable energy community.

Data centers are growing fast, and the way we power them is changing just as quickly. Instead of treating storage as a niche add‑on, we will show how it can play a steady, active role in keeping power stable, smoothing out load swings for the grid or onsite generation, and helping facilities ride through disturbances without disruption.In this session, Devin Dilley, President & Chief Product Officer at EPC Power will introduce power conversion solutions designed specifically for data center applications. As more projects use onsite generation as bridging power until grid interconnection becomes available, our system helps dampen load oscillations when operating alongside synchronous generators.

Accelerated data center deployment often creates a web of complexity, with numerous siloed control systems managing individual processes.Join Phil Jones from Emerson to discover how to meet availability requirements with low-risk, unified technologies that go beyond traditional distributed control: embedded AI capabilities, advanced applications, simulation tools, and cybersecurity solutions, supported by comprehensive lifecycle programs, optimize generation, distribution, and consumption across data center operations.

As AI reshapes every layer of digital infrastructure, data center operations are entering a new era of intelligence. Automation, analytics, and AI-driven monitoring are enabling faster decisions, tighter reliability, and leaner maintenance - all while power, density, and complexity continue to rise. But intelligent operations aren’t built overnight. They require new tools, new thinking, and a workforce ready to collaborate with machines as much as maintain them. This panel explores how operators are re-engineering the modern data center to be smarter, faster and safer for the AI age.This panel will explore:How AI and automation are redefining decision-making and maintenanceBuilding resilience through predictive analytics and real-time monitoringUsing digital twins and unified data for full-site visibilityRedesigning teams and skills for AI-augmented operations

Where AI training workloads demanded the absolute maximum power, and for sustained time periods, the inference shift is driving dramatic rack-level changes. From decreased latency for real-time applications to a lower TCO, the densification of rack-scale architecture and the development of purpose-built inference racks is allowing for optimization of workload performance and power consumption. Join Harsha Bojja, AI Platform Architect at Google, as he explains why and how rack densification is proceeding for inference, and what this means for the wider digital infrastructure ecosystem from inside the data center and beyond.

AI isn’t just changing rack density, it’s exposing weaknesses in how we design resilience.Are we still engineering batteries for yesterday’s data center? Lithium, particularly LFP, is rapidly entering mission-critical environments. Yet hesitation remains: fire risk headlines, AHJ resistance, cost concerns, environmental scrutiny.In this session, MPINarada will challenge common industry thinking and unpack what actually determines whether lithium is risky, or resilient.Why first-cost comparisons are misleading in AI-era facilities?What separates engineered lithium from headline risk?How AHJ approvals are secured and why some projects stall?Where lithium truly outperforms VRLA and where it doesn’tHow energy storage is shifting from backup insurance to strategic power asset

Waiting for utility transmission upgrades and multi-year interconnection queues, the traditional model for data center energization, is no longer compatible with the breakneck speed of AI-driven demand. To stay competitive, developers must pivot from a "passive ratepayer" mindset to an "active power producer" strategy.Join Carson Kearl to explore how to bypass traditional grid constraints by leveraging distributed generation and high-level policy arbitrage. We will break down specific, underutilized utility provisions and state-level regulatory "fast-tracks" that allow developers to energize sites in months rather than years

As AI training clusters scale to hundreds and thousands of GPUs, their economic impact grows but so does their fragility. In tightly coupled distributed workloads, a single link flap, GPU fault, or control-plane inconsistency can cascade into full job termination, forcing costly checkpoint restarts and wasting valuable GPU compute cycles. Traditional approaches treat reliability as a hardware problem and accept failure as inevitable. This talk presents a different model: resiliency and performance as software-defined properties of the cluster. We will examine the real-world failure modes that emerge at fleet scale such as Xid errors, NCCL timeouts, GPU throttling, straggler amplification, fabric contention. We will examine how these localized disruptions propagate across distributed training jobs. From there, we introduce the concept of Software-Driven Fabrics: an architectural approach where observability, communication-layer intelligence, and control-plane coordination work together to detect, isolate, and remediate faults in real time. Rather than restarting the entire training job, faults are isolated at the granularity of individual GPUs, communication links, or ranks. This enables targeted recovery while the remainder of the cluster continues training uninterrupted. We will also explore how these principles become critical in next-generation compute dense systems such as NVL72-based architectures, where tightly integrated fabrics and highly dense compute amplify the effects of these small disruptions. Attendees will gain a practical framework for building AI fleets that operate through failures while sustaining high GPU utilization under real-world conditions, and evolve from fragile batch systems into adaptive, self-healing distributed platforms. The future of large-scale AI training is not just bigger clusters but also intelligent, self-healing, software-driven fabrics that keep training jobs running like clockwork.

Institutions face challenges managing and querying trillions of files/objects across heterogeneous storage systems (e.g., NFS, SMB, parallel FS, archives, cloud buckets) while strictly honoring access controls. Traditional methods make searches slow or impossible, especially for users with access to millions or billions of files. Migrating data to centralized data lakes risks losing embedded access controls, exposes sensitive information, and is impractical for voluminous legacy data. What if one could scale trillions of items, enable fast queries (from weeks to minutes), and integrate with AI for RAG and agentic workflows, turning distributed holdings into a massive, access-controlled knowledge base. Join Gary Grider, Head of HPC Division at Los Alamos National Laboratory, as he breaks down how to turn your data holdings, no matter where they are or what storage system them reside in, into findable knowledge, and what this means for you, whether you're a supercomputing lab, a hyperscaler, or an enterprise.

Speed and sequence in today's market is critical, but not while sacrificing reliability and efficiency. This dilemma is forcing Data Centers and Hyperscalers to solve challenges with a solution: BYOG (bring your own generation), however actual results may vary.Join Dave Bosco, VP at PROPWR, to discuss how an infrastructure grade behind the meter (BTM) power experience is needed to rethink the grid. Learn about strategies PROPWR is implementing, focused on hybrid generation to deliver electrons in bridge, long-term, temporary and permanent generation infrastructure.

AI data centers are redefining power requirements, pushing developers and operators to rethink how speed, resiliency, and long-term flexibility are achieved behind the meter.Join Haroon Inam, CEO and Co-founder of DG Matrix, to explore how multi-port solid-state-transformer solutions enable a standardized, globally deployable power architecture that adapts across diverse site types and grid conditions.Attendees will gain insight into how next-generation behind-the-meter solutions can unlock scalability, resiliency, and grid compatibility for the next wave of AI datacenter growth.

Close out your time with us over open-bar networking drinks on the show floor, courtesy of our hosts Aligned Data Centers, while you enjoy the final content sessions of the day.

The North American construction industry is navigating its most turbulent phase in years, with demand outpacing supply chains, skilled labor, and grid availability. Tariffs, material costs, and regional permitting delays are compounding delivery risks, while investors and customers continue to expect faster timelines and flawless execution. Building smarter, not just faster, has become the new benchmark. This panel examines whether today’s delivery model can sustain the pace of AI growth - and what needs to change to deliver capacity efficiently, compliantly, and at scale.This panel will discuss:How procurement strategies are evolving in the face of tariffs and supply volatilityThe role of modular and standardized design in delivering at speedDelivering density: What are the additional construction considerations?How can we navigate the labor shortage amidst the AI boom?Horizon scanning for supply chain shocks, and how to prepare

AI data centers are scaling to unprecedented gigawatt levels, introducing extreme dynamic loads from GPU synchronization and rapid power ramping. These fluctuations can destabilize grids and generators without advanced power infrastructure.Explore how next-generation lithium-ion batteries absorb high pulse loads, enable load balancing, and stabilize power systems, making batteries a critical component of AI data center power architecture.

Thank you for joining us at DCD>Connect | New York 2026! We hope you had a fantastic time and found it to be a productive experience. We look forward to seeing you again next year!