DigitalBridge and ArcLight Merge in $100B Power-AI Infrastructure Play

DigitalBridge Group and ArcLight Capital Partners announced Tuesday they're combining forces to create what they're calling a leading alternative asset manager positioned at the convergence of power infrastructure and AI-driven digital assets. The merged entity will oversee more than $100 billion in assets under management, betting that the explosive energy demands of artificial intelligence have permanently fused two previously distinct investment categories.

The deal pairs DigitalBridge's decade-long specialization in data centers, fiber networks, and telecom towers with ArcLight's 23-year track record financing power plants, renewable energy projects, and transmission infrastructure. According to the companies, the combination isn't just about scale — it's a structural acknowledgment that hyperscale computing and electricity generation are now inseparable.

Marc Ganzi, CEO of DigitalBridge, framed the merger as a response to what he called "the defining infrastructure challenge of our time." In a statement, he noted that AI workloads are creating unprecedented electricity demands that traditional grid infrastructure wasn't designed to handle. "You can't build a data center anymore without solving for power first," Ganzi said. "This combination gives us the full stack — from generation to distribution to the racks themselves."

The timing reflects a broader reckoning across the infrastructure investing world. Data center operators are increasingly forced to co-locate with power sources or secure dedicated generation capacity just to get projects permitted. Meanwhile, utilities and independent power producers are negotiating directly with hyperscalers in ways that bypass traditional rate structures entirely. The line between digital infrastructure and energy infrastructure is blurring fast — and this merger is a $100 billion acknowledgment of that fact.

DigitalBridge has spent the last decade assembling one of the largest portfolios of digital infrastructure assets globally. The firm manages investments across data centers, cell towers, fiber optic networks, and small cell systems — the physical layer that enables cloud computing, 5G networks, and increasingly, AI model training and inference. The company took itself private in 2024 after years as a publicly traded REIT, giving it more flexibility to pursue long-duration infrastructure plays without quarterly earnings pressures.

ArcLight, by contrast, has been a quiet giant in energy infrastructure since its founding in 2001. The firm has raised more than $30 billion across multiple funds, investing in natural gas pipelines, power generation facilities, renewable energy projects, and midstream assets. ArcLight's portfolio includes stakes in everything from combined-cycle gas plants to solar farms to energy storage systems — the unglamorous but essential infrastructure that keeps electricity flowing.

What neither firm had on its own was full visibility into the convergence happening in real time. DigitalBridge's data center clients were increasingly asking about power procurement strategies, co-location with generation assets, and whether the firm could help finance captive renewable capacity. ArcLight's energy portfolio companies, meanwhile, were fielding inbound interest from hyperscalers looking to lock in dedicated capacity for AI training clusters.

The strategic logic is straightforward: combine the platforms, and you can underwrite an integrated asset — a data center campus co-located with dedicated solar, wind, or natural gas generation, potentially including battery storage to smooth intermittency. That's a product neither firm could credibly offer alone, but one that's becoming table stakes in the AI infrastructure arms race.

The catalyst for this merger isn't subtle. Training a single large language model can consume as much electricity as several thousand homes use in a year. Inference — running those models at scale — is even more energy-intensive when deployed across millions of users. Goldman Sachs estimated in 2024 that data center power demand could grow 160% by 2030, driven almost entirely by AI workloads.

That growth is already straining grids in key markets. Northern Virginia, home to the largest concentration of data centers in the world, has seen utilities implement waitlists for new connections. In Ireland, data centers now consume nearly 20% of the country's total electricity — prompting regulators to effectively halt new data center construction until grid capacity catches up.

Hyperscalers are responding by vertical integration. Google, Microsoft, and Amazon have all signed power purchase agreements directly with renewable developers, bypassing traditional utility procurement entirely. Some are going further — exploring small modular nuclear reactors, investing in geothermal startups, or even building their own natural gas peaker plants adjacent to campus sites.

The DigitalBridge-ArcLight combination positions the merged firm to play multiple roles in these deals — as equity investor in the data center, debt or equity provider to the power project, and potentially as operator or asset manager for integrated campuses. That's a pitch no pure-play digital infrastructure or energy-only firm can make.

While the marquee AI labs and cloud giants are the obvious customers, the real volume opportunity may be in the next tier down. Enterprise AI adoption is accelerating, and companies that want to run proprietary models or sensitive workloads increasingly prefer dedicated or hybrid infrastructure over public cloud. That's creating demand for smaller-scale, regionally distributed data centers — each of which still needs reliable, cost-effective power.

The companies didn't disclose financial terms, which likely means this is structured as a merger of equals rather than an outright acquisition. Both brands will reportedly remain in market during a transition period, though it's unclear how long that lasts or what the unified entity will ultimately be called.

Marc Ganzi will serve as CEO of the combined firm. ArcLight's founder and managing partner, Daniel Revers, will take a senior strategic role, though his exact title wasn't specified in the announcement. The integration will reportedly maintain both firms' existing investment teams and strategies in the near term, with plans to launch new co-investment vehicles that span both asset classes.

DigitalBridge brings approximately $65 billion in AUM, concentrated in digital infrastructure. That includes stakes in Vantage Data Centers, Vertical Bridge (a major U.S. tower company), and Zayo Group's fiber assets. The firm also has significant exposure to edge computing and small cell networks — infrastructure that's increasingly relevant as AI inference moves closer to end users.

ArcLight contributes roughly $35 billion in energy-focused AUM. The firm's portfolio skews toward natural gas infrastructure and renewable generation, with recent emphasis on energy storage and grid modernization projects. ArcLight has also been an active investor in the energy transition, backing developers building utility-scale solar and offshore wind projects.

Together, the firms believe they can offer limited partners a differentiated product: a single fund vehicle that invests across the power-to-compute value chain, capturing returns at multiple layers of the stack.

Infrastructure investing has historically been organized into vertical silos — digital, energy, transport, social infrastructure — with specialist firms dominating each category. Brookfield and Blackstone have broad platforms, but even they tend to run energy and digital strategies as separate businesses with separate teams.

The DigitalBridge-ArcLight merger is a bet that AI-driven convergence is strong enough to justify breaking that model. Whether LPs agree — and whether the operational integration can actually deliver on the thesis — will determine if this becomes a template or a cautionary tale.

For limited partners, this deal presents both opportunity and complexity. On one hand, a combined platform offers one-stop access to what's arguably the highest-conviction infrastructure theme of the decade. On the other, it requires LP investment committees to underwrite both digital and energy risk in a single allocation — which may not map cleanly to existing mandate structures.

Pension funds and sovereign wealth funds often have separate buckets for digital infrastructure and energy infrastructure, managed by different teams with different risk appetites. A blended fund could complicate allocations, especially if the vehicle invests in early-stage renewable projects (higher risk, longer payback) alongside stabilized data center portfolios (lower risk, steady cash flow).

The firms will also need to convince LPs that the integration creates value beyond just co-marketing. The pitch is operational synergy — better underwriting, faster execution, proprietary deal flow from being able to solve both sides of the power-compute equation simultaneously. That's compelling in theory. Proving it in practice will require demonstrating that the combined platform can win deals or generate returns that neither firm could have accessed independently.

This merger doesn't happen in a vacuum. Infrastructure capital has been flooding into AI-adjacent assets for the past 18 months, and competition for quality deal flow is intense.

Blackstone, KKR, and Brookfield have all raised or deployed billions targeting data centers and power infrastructure. Starwood Capital launched a dedicated AI infrastructure fund in 2024. Even traditional real estate investors are pivoting, rebranding industrial portfolios as "AI-ready" logistics hubs or converting telecom hotels into edge computing sites.

Valuations have responded accordingly. Data center pricing in primary markets like Northern Virginia and Phoenix is up 30-40% from pre-AI levels, according to CBRE. Power purchase agreement pricing for dedicated renewable capacity has tightened as hyperscalers compete for the same limited pipeline of shovel-ready projects.

The risk is that capital is moving faster than actual demand. AI's energy appetite is real, but it's still concentrated among a small number of hyperscalers and frontier labs. If AI investment cycles prove more volatile than infrastructure investors expect — or if energy efficiency improvements reduce per-workload power consumption faster than workload volume grows — there could be overbuilding.

DigitalBridge and ArcLight are betting the opposite: that AI is in the early innings of a multi-decade infrastructure build-out, and that the firms positioned to deliver integrated solutions will capture outsized returns. That thesis has been right so far. Whether it stays right through the next cycle is the question.

Building integrated power-and-compute projects means navigating two entirely separate regulatory regimes simultaneously. Data centers face local zoning, environmental review, and often community opposition over noise, water usage, and traffic. Power projects — especially renewable or natural gas generation — face federal permitting (for interconnection), state utility commission oversight, and environmental compliance at multiple levels.

Co-locating the two doesn't eliminate those hurdles. It compounds them. A data center developer that wants to build a dedicated solar farm needs to clear both the data center's environmental impact statement and the solar project's interconnection queue with the local utility or RTO. If the project includes battery storage, that's another layer of fire safety and grid compliance review.

The merged DigitalBridge-ArcLight will presumably have deeper benches for managing this complexity, but that's also a bet that regulatory coordination improves. Right now, the U.S. doesn't have a streamlined framework for hybrid power-compute projects. If anything, the trend is toward more scrutiny as communities push back on the cumulative impacts of data center growth.

Texas, with its deregulated electricity market and business-friendly permitting environment, has become a magnet for integrated projects. Several hyperscalers have announced plans to build data center campuses co-located with natural gas generation or battery storage in ERCOT territory, taking advantage of rules that allow direct procurement outside the traditional utility structure.

Other states — particularly in the Southeast and Midwest — are still figuring out how to handle these hybrid projects. Some utilities see them as a threat to the traditional rate base model. Others see them as a solution to growing industrial load without straining residential customers.

If this merger proves successful, expect copycats. The infrastructure asset management industry has historically been slow to consolidate, with firms preferring to stay specialized and defend their niches. But AI's convergence pressure may force more vertical integration — or at least more strategic partnerships between digital and energy-focused managers.

For LPs, the challenge is deciding whether to follow the convergence or stay diversified across specialists. A combined DigitalBridge-ArcLight fund offers thematic exposure, but it's also a concentrated bet on one firm's ability to execute across two complex asset classes. Some LPs may prefer to allocate separately to best-in-class digital and energy managers rather than trust a newly merged platform to deliver on both.

There's also the question of fund structure. Will the combined firm offer traditional closed-end funds, or move toward open-ended perpetual vehicles that allow more flexible capital deployment? Infrastructure LPs are increasingly asking for perpetual structures that better match the long-duration nature of the assets. A merged platform with $100 billion in AUM has the scale to offer that — but it also introduces liquidity management complexity if too many LPs want out at once.

Step back, and this merger is one data point in a larger realignment. AI isn't just another workload — it's a forcing function that's reshaping how infrastructure gets built, financed, and operated. Data centers are becoming power customers on par with industrial manufacturers. Renewable developers are pitching directly to tech companies instead of utilities. Grid operators are rethinking interconnection rules to accommodate gigawatt-scale loads in markets that weren't designed for them.

The firms that recognize this early — and restructure their platforms accordingly — will have an edge. Those that stay siloed risk getting outmaneuvered by competitors who can deliver integrated solutions.

DigitalBridge and ArcLight are making that bet explicitly. Whether it pays off depends on execution, LP appetite, and ultimately whether AI's infrastructure demands grow as fast and as durably as the market currently expects. But the strategic logic is sound: in a world where compute and power are fused, the asset manager that controls both has a structural advantage.

The industry will be watching to see if the integration works — and if it does, who tries to copy it next.