Artificial Intelligence (AI) is no longer a concept of the future—it’s the driving force of today’s digital transformation. As enterprises, hyperscalers, and financial institutions demand more processing power closer to their users and data sources, the need for strategically placed AI infrastructure has never been greater. At NJFX, we believe the next evolution in digital infrastructure starts at the edge—specifically, at the cable landing station (CLS).

Traditionally, cable landing stations served as passive gateways where subsea cables reached terrestrial networks. But NJFX has redefined this model by transforming our CLS into a full-fledged carrier-neutral colocation hub. This paradigm shift now makes it possible to host high-density, AI-ready infrastructure directly at the CLS—bypassing congested metro routes and enabling faster, more resilient data processing.

The Emergence of Hybrid Cable Landing Stations
The industry is witnessing a significant shift with the rise of hybrid cable landing stations. These facilities integrate subsea connectivity with AI processing capabilities, enabling localized compute power at network entry points. This evolution supports the growing demand for distributed inference deployments, moving beyond traditional hyperscaler models. As noted in Total Telecom, hybrid CLSs allow:

• Lower latency, with AI workloads processed closer to the end user
• Greater flexibility for deploying inference at strategic global entry points
• Modular scalability, adapting to evolving infrastructure needs

Total Telecom also emphasizes that hybrid cable landing stations reduce the need for long-haul data transport, cutting latency for AI-driven applications while optimizing bandwidth utilization.

As these shifts take hold, investors and data center operators must rethink long-term strategies. The move away from ultra-large AI campuses suggests that speculative hyperscaler builds may carry greater risk, particularly those requiring extensive power commitments. Instead, mid-sized deployments of 1MW/5MW/20MW and modular colocation models are becoming more attractive investment targets. This approach enables customers to scale their infrastructure in a secure and manageable way, aligning with operational budgets and long-term digital transformation goals.

For enterprises, the ability to leverage regionalized AI infrastructure offers cost advantages while reducing reliance on traditional hyperscaler platforms. GPU-as-a-service providers and enterprise-owned AI deployments are also gaining traction, further diversifying the AI infrastructure landscape.

 

Why Edge AI Begins at the Shoreline
The rise of AI inference workloads demands low latency, high throughput, and proximity to data sources. Hosting AI edge nodes at the CLS offers a unique advantage:

• Proximity to Global Data Flows
  With direct access to subsea cables connecting North America to Europe, South America, and the Caribbean, NJFX is a global crossroads for data—ideal for AI applications relying on real-time international data ingestion.

• Low Latency Performance
  By connecting directly to the cable landing station, NJFX enables diverse network pathways to legacy hubs like Ashburn, NYC, and Miami—or entirely avoids these congestion points through unique routing options. This provides customers with lower latency and highly resilient options for real-time inference, improving performance for AI-driven applications and mission-critical data workloads.

• High-Density Power & Cooling Capabilities
  NJFX’s first floor features a 4.5MW water-cooled data hall, purpose-built for power-intensive AI applications. The second floor offers air-cooled capacity of up to 500kW, with rack densities reaching 40kW—ideal for modular AI deployments.

A New Standard for AI Infrastructure
AI is only as powerful as the infrastructure behind it. At NJFX, we’ve built a facility that scales intelligently and sustainably. Whether supporting GPUs, DSP’s, or other AI accelerators, we offer:

• Modular deployment options for phased rollouts
• Access to 35+ global and terrestrial network operators for diverse, low-latency routing
• A secure, Tier 3 environment designed for uptime and resilience
• Water-based cooling systems to support high density MW deployments

Enabling AI Ecosystems Through Partnership
The future of AI isn’t built alone. NJFX partners with cloud providers, hyperscalers, hardware vendors, and global carriers to offer a collaborative ecosystem. With our strategic location and carrier-neutral design, we empower organizations to deploy purpose-built AI infrastructure—from core training clusters to distributed inference nodes—right at the intersection of global and terrestrial connectivity.

To meet growing AI data demands, Boldyn Networks recently deployed Ciena WaveLogic 6 technology at NJFX, offering wavelength service that is scalable up to 1.6 Tb/s. This investment enables the bandwidth and optical networking capabilities required for AI-intensive workloads, delivering high-capacity, low-latency connectivity directly to NJFX tenants and interconnection partners.

Future Outlook: CLS at the Forefront of Connectivity
As emphasized by SubTel Forum, the transformation of CLS is reshaping the global connectivity landscape. As key enablers of international data traffic, they are becoming essential components of the digital ecosystem. With continued advancements in technology, strategic planning, and a focus on sustainability, CLS are poised to ensure seamless, resilient, and efficient global communications for years to come.

AI innovation doesn’t stop at the data center core—it extends to the edge. As global data flows increase and inference becomes more distributed, cable landing stations are evolving into essential hubs for intelligent infrastructure. NJFX is proud to lead this transformation, empowering the next generation of AI through unmatched connectivity, high-performance colocation, and a truly global reach.

The future of AI infrastructure is fast, intelligent, and strategically located—and it’s already happening at the NJFX Cable Landing Station.

In today’s hyper-connected world, the role of the cable landing station (CLS) is transforming. No longer simply an access point for undersea cables, the CLS is emerging as a critical hub that networks increasingly rely on for primary connectivity. Recent industry analyses and articles reinforce this shift, underscoring how direct interconnection at CLSs drives lower latency, enhanced resiliency, and cost efficiency.

At NJFX, we have always been forward-thinking. As the first carrier-neutral CLS and colocation facility in the U.S., we’ve been empowering companies with direct, secure access to major subsea cables spanning North America, Europe, South America, and the Caribbean. Today, that model is evolving. Networks are beginning to leverage CLSs as their main point of interconnection, while still using other Points of Presence (PoPs) to ensure network diversity and redundancy.

The Shift: CLSs as Core Hubs, Not Just Edge Assets

Traditionally, network traffic would route through CLSs as a mere stepping stone before reaching larger interconnection hubs like Ashburn or Manhattan. This conventional approach often resulted in increased costs and potential bottlenecks due to long-haul backhauls. However, as noted in a recent article from TeleGeography, networks are now finding strategic advantages in establishing direct interconnection at CLSs. This direct approach not only reduces latency but also minimizes points of failure, supporting the accelerated needs of cloud and AI applications.

Complementing CLSs with Diverse PoPs

While many networks are turning to CLSs as their primary interconnection points, articles in industry publications like Light Reading highlight the value of maintaining a diverse portfolio of connectivity options. By combining the robust capabilities of CLSs with traditional PoPs spread across key markets—such as NYC, Ashburn, Atlanta, and Miami—network operators achieve several benefits:

• Enhanced Redundancy: Routing traffic through multiple, diverse paths minimizes downtime and mitigates the risk of service interruption due to outages or cable cuts.

• Optimized Regional Performance: Direct and localized interconnections support the bandwidth and latency requirements of today’s cloud, edge, and AI applications.

• Flexible, Cost-effective Networks: Diversification permits a tailored approach where each network component is optimized for specific operational needs, ultimately driving down costs.

A comprehensive feature in Data Center Knowledge reinforces this narrative, detailing how hybrid interconnection strategies—integrating both CLS and traditional PoP solutions—are becoming best practice for global network operators who demand resilience and speed.

Why It Matters

For financial institutions, hyperscalers, content distributors, and emerging AI-powered platforms, the advantages of connecting at a CLS like NJFX include:

• Direct Global Market Access: Immediate connectivity to a myriad of subsea cables expands global reach and market opportunities.

• Improved Network Resiliency: The combined power of direct CLS interconnection and diversified PoPs ensures robust network continuity in the face of unexpected challenges.

• Lower Latency for High-Demand Applications: Closer proximity to diverse network nodes minimizes delays, a critical advantage as applications become increasingly time-sensitive.

• Cost-effective Scalability: Reducing the need for extensive backhauling directly translates into economic benefits for expanding network infrastructures.

Looking Ahead

As data volumes skyrocket and real-time applications become the norm, the network industry is undergoing rapid evolution. The direct interconnection at cable landing stations represents more than just a trend—it’s a fundamental shift in network architecture and strategy. New industry insights continue to show that the convergence of subsea connectivity with terrestrial networking infrastructure isn’t just beneficial; it’s necessary to handle the growing demands of an AI-enabled, digital-first global market.

At NJFX, we’re proud to be at the forefront of this evolution. By combining direct CLS connectivity with strategically located PoPs, we’re not just preparing for the future—we’re actively shaping it.

The global economy depends on the silent giants beneath our oceans—subsea cables. These fiber optic lifelines are the unsung heroes of the digital era, transmitting more than 95% of international data, supporting everything from emails to financial transactions and cloud applications. But as with all infrastructure, these cables have a lifecycle, and many are approaching the end of theirs.

Most subsea cables are built with a design lifespan of about 25 years. As these systems age, they face numerous challenges. The optical fibers within the cables degrade over time due to constant exposure to environmental pressures such as temperature changes, water pressure, and natural seismic activity. Electrically powered repeaters, which amplify signals across thousands of kilometers, also wear down and are expensive to replace.

Maintaining these older cables becomes a logistical and financial burden. Repairs often require highly specialized ships and crews, with downtime potentially lasting days or even weeks—impacting critical connectivity. In some cases, the costs of maintenance can rival or exceed the cost of new builds.

From a performance standpoint, older cables were not designed for the data volumes and bandwidth demands we see today. The explosion of AI-driven applications, streaming services, global cloud adoption, and latency-sensitive financial transactions has created a level of network strain that legacy infrastructure simply cannot accommodate. Even with advanced modulation techniques, there’s a physical ceiling to how much capacity can be extracted from these aging systems.

In a world where milliseconds matter and resiliency is non-negotiable, relying on end-of-life infrastructure puts enterprises, governments, and service providers at risk.

Extending Lifespans vs. Building New Systems

An example of extending the lifespan of existing infrastructure is the SEA-ME-WE 4 (South East Asia–Middle East–Western Europe 4) cable system. Originally completed in 2005, this approximately 18,800 km submarine cable connects multiple countries across Asia, the Middle East, and Europe. In March 2024, the SEA-ME-WE 4 consortium completed an upgrade using Ciena’s GeoMesh Extreme submarine network solution. This enhancement increased the system’s capacity from 65 Tbps to 122 Tbps, effectively doubling its previous capacity and extending its operational viability.

Investing in New Systems: Meta’s Project Waterworth

Conversely, companies like Meta are investing in entirely new infrastructure to meet future demands. Meta’s Project Waterworth is an ambitious $10 billion subsea cable system designed to circumnavigate geopolitical tension zones, ensuring secure, high-capacity connectivity. This project exemplifies the proactive approach of building new systems to support the next generation of high-speed connectivity essential for powering AI and digital transformation globally.

These examples illustrate the dual strategies in the industry: upgrading existing cables to extend their service life and constructing new systems to accommodate future demands. Both approaches are vital in ensuring robust and scalable global connectivity.

Real-World Investment in Transatlantic Connectivity

Industry leaders are already investing in the future:

• Google’s Grace Hopper Cable connects the U.S. with the U.K. and Spain. Recently, Colt achieved a milestone 1.2 Tbps per wavelength transmission, reflecting both speed and sustainability gains.
• EXA Infrastructure manages an extensive transatlantic network including systems like Dunant, Havfrue/AEC-2, and Amitié, providing key connectivity between Europe and North America.
• Amazon has filed plans for a new cable linking Ireland to the U.S., adding to its global infrastructure footprint to support cloud services.

According to Telegeography, Between 2023 and 2025, the subsea cable industry is experiencing a significant boom, with investments totaling around $10 billion to bring approximately 78 new systems online, adding over 300,000 kilometers of cable.

These developments underscore the industry’s commitment to expanding and modernizing transatlantic connectivity to meet the growing global demand for data transmission.

Future-Proofing Global Connectivity

Forward-thinking network operators, cloud providers, and infrastructure investors are already responding. New builds are not just about capacity—they’re about preparing for a data-driven future. Cables like Havfrue/AEC-2, Grace Hopper, Dunant, and Amitié are examples of this next wave, offering unprecedented speeds and modern architecture. As older systems retire, these new cables will form the backbone of tomorrow’s digital economy.

The subsea cable industry stands at a crossroads. We can continue to patch aging systems, or we can make strategic investments that enable global progress. With only so many transatlantic crossings possible, now is the time to build the infrastructure that will carry us into the next era of connectivity.

At NJFX, we recognize the importance of this transition and are committed to supporting next-generation infrastructure through carrier-neutral interconnection points and strategic partnerships. As new systems emerge, we’re ensuring that the landing points and ecosystem are ready to support the data demands of the future.

At the recent DatacenterDynamics Connect conference in New York, industry experts convened for an insightful debate on a critical question: “Can New York’s network infrastructure still handle the data surge?” The discussion featured a distinguished panel moderated by Stephen Worn, CTO and Managing Director, NAM, DatacenterDynamics.

Panelists:

• Arthur Valhuerdi, Chief Technology Officer at Hudson Interxchange,
• Jeff Uphues, CEO of DC BLOX
• Phillip Koblence, Co-Founder and Chief Operating Officer at New York Internet (NYI)
• Felix Seda, General Manager at NJFX

This esteemed panel addressed the current state and future readiness of New York’s digital infrastructure amid rapidly increasing data demands, particularly driven by advancements in artificial intelligence, edge computing, and hyperscale expansions.

The Shifting Subsea Cable Landscape

Moderator Stephen Worn opened the discussion by addressing a common industry misconception, asking Felix Seda, “How many people still think that subsea cables land directly in Manhattan or New York City?”

Felix Seda responded, “Historically, yes, subsea cables often landed in New York, but the subsea landscape has dramatically shifted.” He noted that now, cables increasingly land in alternative locations such as New Jersey, Virginia Beach, and Myrtle Beach.

Seda highlighted the critical role of education in correcting these misunderstandings, stating, “We speak regularly with international ISPs who mistakenly believe cables land directly in New York City. One subsea cable lands at NJFX that connects directly to Brazil, bypassing Florida entirely. Educating networks about these physical assets and their true locations is essential.”

Data Surge at the Edge

Phillip Koblence, Co-Founder and COO of NYI, joined the conversation by affirming the centrality of New York in global connectivity despite infrastructure shifts. “The answer to whether New York can handle the data surge is unequivocally yes,” Koblence stated. “Data generated at the edge through technologies like AI, IoT, and inferencing continues to rely heavily on connectivity hubs.”

Arthur Valhuerdi, CTO of Hudson Interxchange, also expressed confidence in New York’s infrastructure readiness, emphasizing high-density capacity expansions: “We’re preparing for significant growth by developing cabinets capable of supporting up to 50kw per rack, specifically to handle emerging inference workloads.” Valhuerdi further noted, “Connectivity requirements within data centers are increasing substantially, with racks now needing upwards of 1,028 fiber connections each. This number is expected to rise further, driving continuous enhancements to our infrastructure.”

Jeff Uphues, CEO of DC BLOX, expanded on the conversation by highlighting hyper-scale growth across the Southeast. “We focus on hyper-scale data centers interconnected by long-haul fiber networks,” Uphues stated, citing examples such as a cable landing station in Burlington, South Carolina, connected via 465 miles of fiber to Atlanta.

“Three subsea cables currently land in Myrtle Beach, with two more planned,” Uphues said. “One of these cables, a 500-terabit link between Santander, Spain, and Myrtle Beach, illustrates how subsea infrastructure drives regional connectivity.”

Uphues emphasized the escalating demand for infrastructure, noting, “One client requires 180-plus reels of fiber per megawatt. We’re building facilities with capacities exceeding 200 megawatts each, illustrating the massive scale and fiber density now demanded in our designs.”

He concluded by underscoring the interdependence of regional growth, adding, “The more infrastructure we build in the Southeast, the greater the need for mature connectivity to major hubs like New York City. This symbiotic growth underscores that it’s not an either-or situation but a collaborative expansion of network infrastructure.”

Phillip Koblence further elaborated on the industry’s maturity, emphasizing diverse needs within digital infrastructure. “The more the industry grows, the greater the overall need for connectivity,” Koblence explained. “It’s critical to understand that digital infrastructure isn’t one monolith; different applications and facilities require specific infrastructural solutions.”

Worn then posed a question about industry maturity to Koblence, asking, “how prepared are today’s networks to support tomorrow’s data-driven world?”

Infrastructure Maturity

Koblence responded, “It varies greatly across verticals. The New York market benefits from longstanding infrastructure laid decades ago, which continues to support modern internet traffic. Despite newer builds, much of the initial groundwork remains highly relevant due to advancements in fiber optic technologies.”

Jeff Uphues provided insight into infrastructure developments in the Southeast, noting, “Unlike New York, we don’t deal extensively with legacy copper infrastructure. Our facilities and networks are entirely new builds, allowing us to leapfrog many of the legacy challenges.”

He expanded, “I was part of MCI’s local service team that helped build out the early internet infrastructure—OC-2, OC-3, OC-12, OC-48 rings, and the first phase of lighting buildings with fiber. What we’re seeing now is the remaking of those early peering points. Hyperscale customers are heading to places with available land and power, where they can build large-scale campuses. These regional interconnection points are tied back into core data centers. It starts with the network,” he added. “Today, we’re talking about thousands and thousands of fiber slices in a facility. These are connected to the major hubs and are remapping the backbone of the internet.”

Felix Seda pointing to a notable shift in enterprise behavior, “Hyperscalers saw this coming. Enterprise customers were behind. They used to be fine letting a backup provider manage their capacity. Now they want control. They want to understand how their network connects directly off the subsea cable and how they can physically route that data to places like Virginia, Chicago, or New York City.”

Seda emphasized, “It’s about having physical overlay and true visibility into the routes their traffic takes. That shift in mindset is driving new infrastructure strategies for enterprise customers across the board.”

Building for the Next Wave

As the panel neared its conclusion, Worn asked a critical forward-looking question: “What’s next? How are you preparing for the next wave, especially in light of unexpected shifts like the Somerset, New Jersey explosion following the stock exchange’s move?”

Felix Seda responded with a focus on control and resilience. “It’s about mitigating risk,” he said. “At NJFX, we’re building infrastructure that increases bandwidth while eliminating middlemen. We’ve acquired the four physical beach conduits from SubCom—owning that last-mile infrastructure gives networks the control they need and avoids chokepoints in legacy hubs like 60 Hudson.”

Seda explained the impact of this ownership model: “By converging physical assets—wavelengths, dark fiber, and direct subsea access—we’re enabling more resilient, congestion-free connectivity paths. Some traffic still needs to go to New York, but increasingly, we’re seeing high-compute workloads move directly from places like Norway into NJFX.”

Phillip Koblence echoed the importance of end-to-end solutions: “We’re trying to get away from calling ourselves just a facility. We’re a facilitator—part of a broader solution. It’s not the customer’s job to figure out how to dig a trench to the beachhead. We’re solving for that so they can focus on what they do best.”

He added a humorous aside: “You can’t sell yourself as ‘New York’s best kept secret.’ That’s why Century 21 went out of business. You need to communicate value clearly—especially when you’re providing critical access like we do with our conduit system at 60 Hudson.”

Koblence also noted their exploration of quantum technologies: “The first wave of quantum commercialization will happen at the network layer—specifically in encryption. And it will rely on dark fiber. You can’t run those workloads on traditional lit networks. We’re laying the groundwork now, even if mass adoption is five years away.”

Ultimately, both leaders emphasized the same principle: supporting innovation by understanding customer pain points and delivering tailored infrastructure that scales with demand. “We’re seeing smarter customers. They know what needs to stay in New York and what can move south to take advantage of power, tax incentives, and real estate. That level of understanding is a big evolution.”

Koblence added, “You can’t separate metro hubs from expansion regions. The growth is mutual and interconnected—both have roles to play in supporting the next wave of data infrastructure demands.”

Final Thoughts: Capacity, Continuity, and Commitment

Arthur Valhuerdi wrapped the session with a focus on short-term readiness: “We’re focused right now on our high-density product—just seeing what kind of customers are coming and where demand is headed. We’ve got seven megawatts still to deploy and may dedicate an entire floor to high-density if that need materializes.”

Even if inference engines don’t fully materialize in the short term, Valhuerdi believes demand is inevitable. “Whether it’s 600kW cabinets or traditional workloads, all of this capacity will be absorbed. The internet isn’t a fad anymore—it’s foundational.”

He underscored that his goal remains simple: meet the needs of customers, whatever their content or application. “Especially at sites like 60 Hudson, 165 Halsey, and One Wilshire—it’s all about connectivity.”

Jeff Uphues shared a long-term view: “Eight years ago, we set out to build a dozen data centers across the Southeast within a decade. We’ve completed or have in development ten so far.”

He emphasized that their mission has remained consistent: “We’re focused on building network-centric infrastructure—interconnecting hyperscale edge nodes and data centers with a robust fiber backbone. We’ll continue to invest in subsea infrastructure and long-haul fiber.”

Reflecting on the company’s progress, Uphues said, “It hasn’t always played out in the exact way or timeline we envisioned, but our focus never wavered. It’s all about where we win, how we best serve our customers, and staying relentlessly committed to our strategy.”

With that, the panel concluded—each expert reinforcing the evolving nature of digital infrastructure and the collaborative responsibility to shape its future

The hype around artificial intelligence (AI) in data centers has never been greater and this panel discussion at the Pacific Telecommunications Council (PTC) Conference discussed how we are transforming the world to drive AI. Industry leaders gathered to tackle the seismic shifts in data center infrastructure brought on by the surge in AI-driven workloads. Attendees were eager to hear how companies are preparing for the next wave of high-performance computing and what it takes to support GPU-driven infrastructure at scale.

Moderated by Michael Elias, Senior Equity Research Analyst at TD Cowen, the panel featured top executives shaping the future of digital infrastructure:

• Chris Sharp, CTO, Digital Realty
• Gil Santaliz, CEO, NJFX
• Jaime Leverton, CEO, Jaime Leverton Ventures & Advisory
• Bjorn Brynjulfsson, CEO, Borealis Data Center

Elias set the stage by directing the first question to Chris Sharp: What is required to run GPU-powered infrastructure, and how is Digital Realty adapting to meet the needs of high-performance computing?

Sharp began by acknowledging the unprecedented scale of AI’s growth and how it is fundamentally changing the way data centers operate. “The quantum’s that we’re talking about today, the momentum, the pacing—it’s astronomical,” he stated. “Digital Realty, currently has 2.7 to 2.8 gigawatts of IT load deployed, with an additional three gigawatts coming online in rapid succession.”

This doubling of capacity underscores the massive infrastructure shift required to support AI as a workload. He attributed this surge to transformers, the AI models behind ChatGPT, which have intensified power density and cooling demands. “Liquid cooling is now essential, with over 50% of Digital Realty’s capacity modular enabling densification up to 150 kW per rack for Nvidia deployments,” said Sharp.

The shift toward AI infrastructure brings additional challenges. AI racks can weigh up to 5,000 pounds, requiring reinforced floor structures to support their density. The demand for AI workloads has also increased cabling requirements by five times, adding to infrastructure complexity and costs. Sharp added, “Network efficiency has become a critical factor, with InfiniBand imposing distance limitations that require precise architectural design.”

“AI infrastructure extends beyond compute power. Connectivity through carrier-neutral facilities and subsea cables is just as crucial in enabling low-latency AI workloads,” Gil Santaliz adding that the critical role of connectivity in AI deployments is essential and secure.

By facilitating high-capacity traffic between subsea and terrestrial networks, NJFX enables AI workloads to function with minimal latency and maximum efficiency. Without this seamless connectivity, even the most advanced AI models risk becoming isolated and ineffective, unable to deliver real-time insights where they are needed most. “Connectivity is key,” Santaliz emphasized. “NJFX sits uniquely between Virginia and Boston, offering direct access to four subsea cables connecting Denmark, Norway, Ireland, and Brazil. This strategic positioning enables us to provide AI inference as a platform for global collaboration.”

Santaliz emphasized that AI inference, which allows trained models to generate rapid real-time data, remains uncertain who is leading the charge. “The population has to get access to these language models that are trained for production,” he said. “We’re going to see smaller, edge-performance five-megawatt deployments in population centers of 100 million people within five milliseconds.”

Deploying AI inference at this scale is not without challenges. “This isn’t a cheap sport,” Santaliz noted. “The cost of generators, transformers, and switchgear has doubled in the last 24 months due to supply chain issues. Bringing in five megawatts of power, plus the infrastructure to support liquid cooling, requires significant investment.”

Despite these hurdles, he believes a successful model will emerge. “Hyperscalers will keep inference within their existing campuses, while others may deploy in the New York metro area. The economics are staggering, but those who do it right will set the standard for the rest of the industry.” Santaliz also pointed to industry leaders like Digital Realty and Equinix as key players in this evolution, but emphasized that more capacity is needed. “The land grab has already happened—so now, where do you put these inference nodes?”

As AI continues to evolve, the panelists agreed that balancing compute power with strategic connectivity and infrastructure investment will be critical to meeting the demands of AI inference at scale.

The moderator transitioned to Jaime Leverton asking how organizations are navigating the infrastructure changes required for AI and high-performance computing. Leverton noted that many companies are seeking expert guidance rather than trying to figure it out in-house. “Things are moving rapidly, it’s not possible to do this on your own unless you’re working day in and day out,” she explained. “I see most organizations looking for external expertise, which I believe is a smart approach.”

Leverton also mentioned the increasing migration of power resources from Bitcoin mining to the data center space. “I was recently the CEO of a large, publicly traded Bitcoin mining company, and I’ve noticed more capital and operators from that sector shifting into traditional data centers,” she said. “It’s an interesting trend that isn’t widely discussed, but it’s shaping the infrastructure landscape.”

Jumping back into the discussion, Gil Santaliz expanded on the importance of power sourcing for AI workloads. “The power of GPU infrastructure is driving workloads closer to the source of energy,” he noted. “Moving data is more cost-effective than moving energy, which is why we’re seeing shifts like workloads migrating from continental Europe to the Nordics, where power is more abundant.”

Bjorn Brynjulfsson added how industries are repurposing existing infrastructure for AI expansion. “In Finland, we’re seeing old paper mills transformed into high-capacity data centers, tapping into hundreds of megawatts of available power. These types of conversions are creating new opportunities,” he explained. “For example, in Iceland, we’ve upgraded older facilities to support 50 kW per rack using direct air cooling while maintaining US standards.”

As AI scales, infrastructure will continue to adapt, blending new and traditional approaches to ensure efficiency and sustainability.

Sharp provided insights into the future of AI-driven infrastructure, emphasizing the need for efficient cooling solutions and modular liquid cooling integration. “Sufficient allocation of capital is what wins,” he stated. “At the core of our business, we have to be mindful of technological obsolescence while ensuring scalability.” No single design works universally, requiring companies to adapt based on location and climate conditions. Sharp cited a DGX deployment in the Nordics, where free air cooling significantly improved power usage effectiveness (PUE). “What’s unique about that facility is the ability to integrate liquid cooling efficiently, something that remains rare at scale,” he added.

Sharp emphasized that AI is not just a technology—it’s transforming every industry. “Enterprises that embrace AI will gain a major competitive edge,” he said. “Those thaat fail to adapt will struggle as AI reshapes workflows and business models. AI’s rise to the SaaS revolution, predicting that enterprises will increasingly buy AI-driven services rather than developing in-house solutions,” he noted.

Sharp also pointed to private AI deployments as a growing trend, with companies leveraging AI to gain deeper insights from their vast datasets. “Hyperscalers are leading in GPU deployments, but enterprises are realizing the importance of controlling their own data,” he explained. “We’ve moved from data lakes to data oceans, and having access to those insights is a major differentiator.”

With AI models becoming more sophisticated, the discussion underscored the importance of strategic cooling, efficient data management, and long-term infrastructure investments to support AI’s rapid expansion.

Shifting to Brynjulfsson, as the moderator asked how data centers can plan for a 20-year lifespan amid accelerating rack densities. He acknowledged the challenge of making long-term infrastructure decisions in a rapidly changing environment. “It’s tricky. You’re making investments with long-term impacts, and we’re currently building a PLC solution that’s auto-collect in size, targeting 130 kW on the DLC side,” he explained. “However people are already talking about 250 kW per rack, so we have to think ahead.”

When asked if planning for a 20-year timeframe is even realistic, Bjorn acknowledged that some infrastructure will remain, but upgrades will be inevitable. “We previously optimized for blockchain in 2019, using high power in a small footprint,” he shared. “But we knew we’d transition to HPC eventually, so we made our facilities upgradeable. We didn’t anticipate going beyond 150 kW per rack at the time, but we’ve had to add more busbars and power feeds.”

Bjorn noted it may make more sense to build from scratch rather than continually upgrading existing facilities, making planning a crucial part of the data center evolution strategy.

A critical issue emerged—the strain on utilities to meet the surging power demands for AI workloads. The industry is facing a paradox: while there is a significant land grab for AI deployments, the reality is that power availability is a bottleneck. This artificial demand risks stalling genuine, scalable AI infrastructure projects.

The key takeaway? If you have power and an operational facility, you’re in a strong position. Securing generators, transformers, and utility connections has become a multi-year process, and companies that fail to plan ahead may find themselves unable to execute AI deployments on schedule.

The industry must balance power constraints, infrastructure preparedness, and software optimizations to enable AI’s next phase. The future will likely see greater integration between AI inference and training models, shifting workloads dynamically based on available capacity.

“2024 has been a year of rapid land acquisitions, but 2025 will be defined by efficiencies—determining where AI inference should live and how it should scale,” Sharp added.

Panelists emphasized that AI inference is still in its infancy, and its architecture is evolving rapidly. No two models operate the same way, and as inference workloads become more specialized, they will require tailored infrastructure solutions. AI’s future will not be a one-size-fits-all model but rather a mix of edge inference, centralized processing, and optimized power distribution.

 

 

 

December 11, 2024 – Capacity CALA –  The panel discussion focused on the the critical intersection of technology, governance, and resilience in the face of natural disasters affecting the Caribbean/LatAm region. Moderated by Gil Santaliz, CEO and Founder of NJFX, the panel brought together industry leaders to discuss strategies for disaster preparedness, the role of technology in mitigation, and the essential partnerships between governments and private enterprises.

Introducing the Panelists

• Gil Santaliz, NJFX CEO and Moderator
• Andy Bax, Senior Partner for Digital Infrastructure at Cambridge Management Consulting
• Teresa Wankin, Secretary General of CANTO
• Matt Fouch, President and Co-Founder of Subsea Data Systems

Building Resiliency in the Caribbean: Technology, Collaboration, and Government Incentives

The Caribbean is no stranger to natural disasters. Each year, hurricanes, tropical storms, and other extreme weather events threaten the region’s critical digital infrastructure. At a recent discussion held at the PGA National Resort, industry leaders, government representatives, and stakeholders gathered to address the urgent need to strengthen connectivity and disaster preparedness. Moderated by Gil Santaliz, CEO and Founder of NJFX, the conversation drew on decades of collective experience to outline a path forward, combining innovative technology solutions with strategic policy and inter-regional collaboration.

“The Caribbean faces unique challenges in maintaining robust digital infrastructure amidst frequent natural disasters,” said Teresa Wankin, Secretary General of CANTO. She recalled the devastation created by Hurricane Maria, “The entire region felt the brunt of Maria, and in Puerto Rico, there was considerable damage to the landing stations, which left them out of connectivity for quite some time.”

Wankin emphasized that business continuity planning cannot be an afterthought. “We’ve seen time and time again the importance of business continuity planning, pre, during, and after,” she explained, highlighting CANTO’s efforts in uniting operators, cable companies, and regulators. “We worked with a private company to develop a framework for how to treat disasters. The importance of multiple landing stations to mitigate volcanic activity and regulatory challenges cannot be overstated within the region.”

Diversifying Infrastructure
Andy Bax, a Senior Partner at Cambridge Management Consulting with over 30 years of experience in building and operating submarine cable systems, stressed the necessity of resilient network architectures. “The economic model of diversifying landing sites and collaborating with South American networks can significantly reduce reliance on single points of failure,” he said. Sharing insights from his time in Indonesia after the tsunami, Andy underscored the importance of communication when catastrophe strikes. “When you’re in a disaster zone, the ability to communicate is absolutely critical.”

Andy’s perspective shed light on both the technological and policy components of resilience. “We architect networks with a plan for failure, ensuring that if one segment goes down, others maintain connectivity. It’s not just about technology—it’s about planning, outreach, and making sure everyone, from NGOs to local operators, knows how to respond.”

Turning to the future of infrastructure resilience, Matt Fouch, President and Co-Founder of Subsea Data Systems, discussed the technological advancements that are being place on wet cables, such as SMART Cables. These subsea cables would integrate sensors to monitor environmental and operational parameters in real time. “Smart cables are intended to be sensors connected to the cable, providing real-time data on seismic activity, pressure, and temperature,” said Fouch. “With smart cables, we can sense potential cable breaks before they happen, determine the cause, and respond proactively.”

Cost remains a barrier. “Our goal is to reduce the cost of sensor systems from $1.5 million to around $150,000, making it feasible to deploy smart cables across the Caribbean and Latin America,” Fouch explained. He acknowledged the economic hurdles but insisted the investment would pay dividends in disaster preparedness, “Funding remains a challenge, but the potential impact on resilience and safety is immense.”

Bridging the Economic Gap and Enabling Action Through Cooperation
Technology alone cannot solve these challenges if operators, governments, and businesses are not incentivized to implement it. Both Teresa and Andy agreed on the importance of government involvement. “Governments need to incentivize operators through mechanisms like the Universal Service Fund to invest in resilient infrastructure,” suggested Theresa. Andy concurred, noting that without financial support and a supportive regulatory environment, companies might resist investing in long-term projects that do not yield immediate returns.

Santaliz pointed out a case in Puerto Rico, where a state-of-the-art facility remained underutilized because operators could not justify the expense of relocating their equipment. “We have these beautiful, hurricane-hardened facilities, but without incentives, operators stay put,” Santaliz observed. This illustrates the need for public policies and partnerships that encourage migration into safer, more resilient infrastructure and networks.

Fouch underscored that the responsibility for funding and implementing such technologies should not fall solely on island nations. “Governments have funded nearly all the science that helps us understand climate impacts,” he said. “They need to step in here as well.” International agencies, including the United Nations, could play a role in coordinating early warning systems and sharing best practices across borders.

The panelists consistently returned to the theme of cooperation—between governments and industry, operators and regulators, and even across regions. “This is about finding ways to work together so that when disaster strikes, we’re not playing catch-up,” said Wankin. “We need a framework that allows everyone to respond quickly, effectively, and collectively.”

A Shared Responsibility
As the discussion wrapped up, the consensus was clear: building a more resilient infrastructure requires a multifaceted approach. It demands strategic investments in technology like smart cables, sensible policies and economic incentives from governments, and a sustained commitment to collaboration at every level.

“Collaboration with regulators and governments is essential to create an environment conducive to investment and resilient infrastructure,” Wankin reaffirmed. Fouch underscored the transformative potential of sensor-based networks, “The goal is to make these technologies accessible, so every island can monitor its environment and anticipate challenges before they become crises.”

Bax concluded by calling for continuous outreach and enforcement: “We must use available technologies to enforce protection measures and maintain outreach programs to safeguard our critical assets.”

The collective insights from the panel at Capacity CALA reflect a vision of a more secure, connected, and resilient Caribbean. By embracing innovation, fostering public-private partnerships, and ensuring that policies, incentives, and international cooperation are in place, the region can protect its infrastructure, reduce downtime during disasters, and ultimately safeguard the well-being and economic stability of its communities.

Key Takeaways:

• Collaboration is Crucial: Effective disaster preparedness requires seamless cooperation between operators, regulators, governments, and private enterprises.
• Embrace Technological Innovations: Smart cables with integrated sensors can provide real-time data and early warnings, significantly enhancing resilience.
• Diversify Infrastructure: Multiple and strategically located landing points ensure continuity of connectivity even when parts of the network are compromised.
• Government Incentives Needed: Public-private partnerships and supportive regulatory frameworks are essential to encourage investments in resilient infrastructure.
• Address Man-Made Risks: Developing protocols and enforcement mechanisms can mitigate the impact of intentional disruptions to critical communication networks.

The rapid expansion of data center capacity and the growing demands of AI were at the forefront of a recent executive panel discussion that brought together leaders from NVIDIA, CBRE, Supermicro, EXA Infrastructure, Bulk and NJFX. The panel explored critical issues shaping the industry now with AI’s transformative impact to emerging energy solutions for powering future infrastructure.

Stuart Dyer, First VP-Data Center Advisory at CBRE, outlined the unprecedented growth of data center capacity, driven by hyperscalers like AWS, Microsoft, and Google. By 2026, Northern Virginia alone will account for nearly a gigawatt of capacity, with over 75% pre-leased. The vacancy rate across North America remains a mere 1% underscoring the intense race for power and space.

In Europe, secondary markets such as the Nordics are emerging as key players, thanks to their focus on green energy. With the AI boom accelerating, these regions are becoming hotbeds for hyperscale tenants seeking sustainable and scalable solutions.

AI’s Influence on Data Center Design

Malcolm deMayo, VP Global Financial Services Trusted Advisor at NVIDIA, highlighted the critical role of data centers in supporting AI applications, such as weather prediction and insurance modeling. With Nvidia commanding 88% of the AI market, the need for low latency and efficient connectivity is paramount. The discussion emphasized the importance of subsea cables and scalable networks to support the ever-growing data transfer demands of AI inference and training.

Liquid cooling emerged as a key innovation for managing high-power density racks, with Super Micro showcasing advanced solutions like cold plates and liquid-cooled racks. These technologies not only enhance performance but also significantly reduce power consumption.

Advanced Cooling Solutions

Michael Watson, Senior Director, field Application Engineer at Super Micro, highlighted the company’s global presence and leadership in liquid cooling solutions for high-performance computing and AI. He outlined the challenges of managing heat in modern computing systems, emphasizing the need for innovative approaches to meet the demands of increasingly dense workloads.

Watson detailed SuperMicro’s advanced product portfolio, including cold plates, cooling distribution units, and liquid-cooled racks, all designed to optimize cooling efficiency. These solutions significantly reduce power consumption while boosting performance, making them essential for AI-driven environments. Watson underscored the transformative impact of liquid cooling on the data center industry, offering scalable, efficient systems to support the growing demands of high-density AI workloads.

A Path to Sustainability

The conversation shifted to the future of power generation, with Hunter Newby addressing the pressing need for alternative energy solutions. Small cell nuclear reactors were identified as a potential game-changer, offering scalable and localized power sources for data centers. However, the 5-10 year timeline for regulatory approval poses challenges.

Panelists critiqued the inefficiencies of utility-based power systems, which often struggle with supply chain delays and outdated infrastructure. Hyperscalers like Google were urged to consider building their own power plants to bypass regulatory bottlenecks and meet immediate energy demands.

Kristian Kofoed-Solheim of Bulk Infrastructure highlighted Norway’s green energy advantage, leveraging hydropower to fuel sustainable data center operations. With key points discussing the Havfrue cable, the only cable connecting the United States to the Nordics providing access to these power sources. His remarks underscored Europe’s shifting stance on nuclear power and the growing reliance on renewable energy.

AI’s Impact on Network Design

Charles Thomas, VP Sales Engineer from EXA Infrastructure, highlighted the growing demand for high-bandwidth connectivity to support AI workloads. He outlined challenges in managing existing subsea systems and emphasized the need for software-defined networking to extend network capacity. Collaboration with partners and innovation in subsea systems are critical for handling the surge in data transfer demands fueled by AI.

Thomas also discussed the role of government funding in infrastructure projects, which can accelerate innovation. He emphasized how partnerships between private companies and public entities are essential for building and operating scalable data center solutions.

AI, Connectivity, and Partnerships

Gil Santaliz emphasized NJFX’s pivotal role as North America’s first carrier-neutral cable landing station, crucial for interconnecting continents and supporting AI applications. He highlighted NJFX’s infrastructure, which supports high-density data centers and seamless subsea cable connectivity essential for AI’s low-latency and high-bandwidth needs.

Santaliz pointed out the inefficiencies in traditional utility systems, such as equipment delays and domestic manufacturing constraints. He proposed nuclear power as a solution for data centers to bypass these utility challenges and achieve scalable energy generation.

Additionally, he discussed the complexities of water management in data centers with liquid cooling systems. NJFX’s flexible design accommodates advanced cooling technologies, meeting the increasing demand for AI and high-performance computing. Santaliz concluded by stressing NJFX’s commitment to adapting infrastructure for the evolving needs of AI-driven industries, ensuring efficiency and scalability.

Key Takeaways

1. AI Drives Growth: Hyperscalers and Fortune 1000 companies are accelerating demand for data center capacity to support AI workloads.
2. Energy Innovation is Critical: The transition to small cell nuclear reactors and green energy solutions is vital for scalable, sustainable growth.
3. Connectivity Matters: Low latency and high-bandwidth networks, powered by subsea cables, are essential for AI inference and real-time applications.
4. Collaboration Fuels Progress: Partnerships across the industry are driving innovation and addressing challenges in infrastructure and energy.

As the data center industry evolves, leaders must embrace innovative technologies and sustainable energy solutions to power the future of AI and digital infrastructure. This session offered a glimpse into the strategies and partnerships shaping the road ahead.