JPMorgan and MUFG Near $22 Billion Data Center Financing Deal in Texas, FT Reports

Executive summary

JPMorgan Chase and Mitsubishi UFJ Financial Group (MUFG) are reportedly close to underwriting a colossal $22 billion infrastructure loan to fund Vantage Data Centers’ new hyperscale campus in Texas—a project expected to total about $25 billion, with $3 billion in equity from Vantage’s owners Silver Lake and DigitalBridge. The campus—branded “Frontier”—spans roughly 1,200 acres and is planned to include 10 data centers designed for AI-grade, ultra-high-density compute. The first building is expected to come online around mid-2026, with broader build-out targeting 2028, according to multiple reports. If finalized, this would be one of the largest single data center financings ever arranged, underscoring the ferocious capital needs behind the AI boom. Financial TimesReuters+1

What’s being financed—and why it matters

The “Frontier” campus is envisioned as a next-generation platform purpose-built for AI workloads. Compared with traditional enterprise hosting, AI training and large-scale inference demand orders of magnitude more power per rack, more intricate liquid-cooling solutions, and highly resilient interconnects. Vantage’s Texas site is slated to support ultra-high-density racks north of ~250 kW, positioning it squarely in the top tier of AI-first facilities. In power terms, reporting around the launch has referenced multi-gigawatt ambitions, placing this campus among the largest private-sector compute projects currently in development. Reuters

Beyond sheer size, the financing signals how the AI infrastructure stack is being capitalized. Unlike the early cloud era—when hyperscalers self-funded most expansion—today’s AI rush is bringing banks, infrastructure funds, and private equity into the critical-compute buildout. The $22 billion underwriting led by JPMorgan and MUFG would be a defining example of this shift, with additional lenders expected to join a broader syndicate. Financial TimesReuters

The key players

Vantage Data Centers
Founded in 2010, Vantage has grown into a global developer and operator of hyperscale campuses. Backing from DigitalBridge (which acquired Vantage in 2017) and Silver Lake has enabled rapid expansion and large-ticket, multi-site programs tuned for AI customers. Financial Times

Silver Lake & DigitalBridge (Equity sponsors)
The pair are reportedly contributing $3 billion in equity to the Texas build, complementing the jumbo loan and anchoring long-dated, capex-heavy development with patient sponsor capital. Both sponsors have been active across digital infrastructure—data centers, fiber, and edge assets—where secular demand is strongest. Financial TimesReuters

JPMorgan & MUFG (Lead underwriters)
The banks are said to be near an underwriting for $22 billion of debt—among the largest single financing packages arranged for a data center project. Large underwriting commitments like this often imply confidence that the debt can be syndicated across a broad lender group once terms are finalized. ReutersBloomberg.com

Other banks
Reporting indicates that additional major lenders—such as Goldman Sachs—are part of the wider financing effort, reflecting the scale and syndication required to place multi-tens-of-billions in infrastructure paper. Financial Times

Why Texas?

Texas has rapidly become America’s proving ground for AI-grade compute campuses. There are several reasons:

Power availability and scale
The Electric Reliability Council of Texas (ERCOT) grid offers developers access to large blocks of power and competitive wholesale pricing. While interconnection timelines and transmission bottlenecks can be challenging, developers view Texas as one of the few places where multi-hundred-megawatt phases are feasible on repeat.
Land, permitting, and pace
Large, contiguous land parcels with favorable local permitting environments make Texas attractive for multi-building, multi-year rollouts like Frontier’s proposed 10-data-center plan. Reuters
Pro-development policy climate
The project has drawn support from state leadership—another signal that compute has moved into the category of strategic infrastructure for economic development. Financial Times
Network and proximity
Texas’s central geography facilitates connectivity to both coasts and Latin America, and puts massive AI clusters closer to energy generation and—potentially—new, dedicated renewables.

The financing lens: how a $22B loan might be structured

While specific terms are not yet public, deals of this magnitude typically blend construction facilities, delayed-draw term loans, and possibly infrastructure-style tranches designed for long-lived assets. Underwriting banks commit to fund the package up front and then syndicate pieces to other banks and institutional lenders.

Key features likely to apply:

Staged drawdowns tied to construction milestones across multiple buildings (Frontier has up to ten planned).
Interest rate hedging to manage floating-rate exposure during construction and ramp.
Power and offtake alignment, where anchor customer contracts or capacity reservations help size leverage and price risk.
ESG and efficiency covenants may figure into pricing, especially given cooling innovations and power sourcing strategies for AI loads.
Potential refinancing via term debt or capital markets once the campus stabilizes (leased and energized). Financial TimesReuters

The reported $3B equity from sponsors helps absorb construction risk, smooth draw schedules, and support coverage ratios—critical for a project whose capex profile runs into the tens of billions. Financial Times

The AI-infrastructure super-cycle

The numbers here aren’t happening in a vacuum. Across capital markets, data centers have become one of the most prized real-asset categories, buoyed by cloud migration, streaming, and now AI. The incremental step-change is the AI training and inference layer, which requires:

Massive power budgets: AI racks can push >250 kW each, multiples above conventional densities.
Advanced cooling: Liquid cooling and immersion systems move from niche to standard.
High-bandwidth, low-latency interconnect: InfiniBand/Ethernet fabrics and fiber backbones scaled for cluster-level performance.
Supply chain depth: From transformers to chillers to GPUs, developers must secure components with long lead times.

Market commentary has emphasized just how dramatically valuations and capex plans have expanded in the AI era, with data center market capitalization surging since 2019 and the asset class ranking among the top performers in real estate and infrastructure. Reuters

Project milestones and delivery cadence

Early reporting indicates:

Campus size: ~1,200 acres.
Total investment: ≈ $25 billion.
Debt/equity mix: ≈ $22B debt + $3B equity (Silver Lake + DigitalBridge).
Buildings: 10 planned data centers.
First completion: Mid-2026 target for the initial facility; broader rollout toward 2028.
Density: AI-class, >250kW per rack design targets. Financial TimesReuters

As with all mega-campuses, individual buildings will likely be delivered in phases, allowing customers to begin deploying workloads well before the entire campus is complete. Phasing also helps match capital deployment with signed demand.

Strategic implications

For banks and credit markets

If the underwriting proceeds as reported, it would demonstrate that global money-center banks remain willing to marshal tens of billions into single-sponsor, single-campus compute projects—provided there’s a strong sponsor base, line-of-sight to power, and robust customer demand. It also underscores Japan’s MUFG as a heavyweight in cross-border project finance, especially in digital infrastructure. Reuters

For sponsors and operators

For Silver Lake and DigitalBridge, the Texas program is a flagship bet that AI demand will remain structurally strong through the second half of the decade. A campus capable of multi-gigawatt draw implies a target customer set spanning hyperscalers, AI labs, and enterprise AI programs that require dedicated capacity and power commitments. Financial TimesReuters

For AI customers

Frontier-class sites give AI model developers and cloud providers room to scale without the parcel-by-parcel compromises of smaller markets. With lead times on GPUs, transformers, and grid interconnects, having reserved shells and power in a single campus can translate into meaningful speed-to-compute advantages.

For Texas

The state continues to bolster its positioning as a compute corridor, with potential knock-on effects for grid planning, renewable build-outs, and workforce development. These projects often catalyze new power purchase agreements, transmission upgrades, and collaboration with local colleges on technical training.

The risk ledger

Even with sponsor pedigree and bank backing, projects of this scale carry non-trivial risks:

Power procurement and grid constraints
Accessing multi-hundreds of megawatts of reliable, affordable power is table-stakes—and complicated. Queue congestion, transmission upgrades, and generation mix volatility must be actively managed.
Cooling technology and reliability
As rack densities climb, liquid cooling and advanced heat-rejection systems become core infrastructure. Execution risk spans design, construction, and long-term maintainability.
Supply chain and lead times
Transformers, switchgear, chillers, and specialized materials can carry long lead times. Any slippage can cascade across construction phases.
Interest rates and financing markets
A $22B loan will be sensitive to benchmark rates and credit spreads. Hedging can mitigate some exposure, but sustained rate volatility could pressure economics.
Policy and permitting
Texas is generally favorable, but environmental reviews, water usage, and local approvals still matter—especially for cooling and backup generation.
Customer concentration
AI demand is deep but concentrated. If one or two hyperscale anchors reprioritize, leasing cadence could be affected.

The presence of experienced sponsors and top-tier banks helps mitigate these risks, but does not eliminate them. Financial TimesReuters

How this reshapes the data center playbook

A few patterns are crystallizing from deals like Frontier:

Bigger, fewer, denser: Instead of dozens of 20–50 MW sites, developers are gravitating to mega-campuses where they can replicate buildings, standardize parts, and pool power and cooling at industrial scale.
Bank-sponsored scale: The capex envelope for AI campuses is too large for equity alone. Expect bank underwritings and multi-tranche syndications to become common, with infrastructure debt funds participating alongside commercial banks.
From real estate to energy-tech: Data centers are evolving from a niche real-estate specialty into energy-technology platforms. Power procurement, grid interconnection, and thermal engineering now sit at the strategic core.
ESG tension—and opportunity: Massive power draws attract scrutiny, but they also anchor new renewable projects and grid upgrades. Financing terms may increasingly tie to efficiency and clean-power sourcing.

What to watch next

Underwriting terms and syndication
Final pricing, tenor, and covenants will reveal how lenders are underwriting AI-specific risk. Keep an eye on the order book and ultimate distribution.
Power agreements
Announcements around PPAs (renewables and firming resources), on-site generation, or grid capacity reservations will be crucial.
Anchor customers
While specific tenant discussions are rarely public, any disclosure of capacity reservations or design-win partners will help explain the project’s demand profile.
Cooling and efficiency innovations
Watch for immersion and direct-to-chip liquid cooling at industrial scale, plus waste-heat recapture pilots.
Local ecosystem build-out
Training programs, vendor campuses, and secondary services (from fiber to specialized maintenance) typically follow mega-projects of this type.

Frequently asked questions (FAQ)

Is the deal finalized?
Not yet. Multiple outlets report that JPMorgan and MUFG are near an underwriting for the $22B loan, with final details pending and other banks expected to participate. Reuters

Who owns Vantage Data Centers?
Vantage is backed by DigitalBridge (owner) and Silver Lake (long-time partner and investor). For the Texas campus, they’re reportedly contributing $3B in equity. Financial Times

How large is the campus?
Roughly 1,200 acres, with 10 planned data centers and AI-class density (>250 kW per rack), making it one of the most ambitious projects in the current pipeline. Financial TimesReuters

When will it go live?
Reporting suggests the first building aims for mid-2026, with additional phases targeted through 2028. Timelines can shift based on power and supply chain factors. Financial Times

Why is AI driving such huge capex?
Training frontier-scale models and serving global inference require massive, low-latency compute and specialized cooling, elevating both power density and facility cost—hence the need for very large, multi-phase campuses and multi-billion-dollar financing. Reuters

Bottom line

The reported $22 billion underwriting that JPMorgan and MUFG are nearing for Vantage’s Texas “Frontier” campus is more than a single mega-deal; it’s a snapshot of the new AI infrastructure economy. Banks, private equity, and dedicated digital-infra specialists are converging to fund AI-first campuses at unprecedented scale. For Texas, it reinforces the state’s position as a national hub for high-density compute. For sponsors and lenders, it sets a template for how the next generation of AI capacity will be built: fewer, bigger, denser—and financed like critical infrastructure. Financial TimesReuters+1

Reporting sources: Financial Times first detailed that JPMorgan and MUFG are near an underwriting of a $22B loan for Vantage’s Texas campus, with $3B in equity from Silver Lake and DigitalBridge and a 10-building, 1,200-acre plan. Reuters, Bloomberg, and industry trade coverage corroborated key elements and added context on project scale, timelines, and the broader AI data-center investment cycle. Financial TimesReutersBloomberg.com

https://bitsofall.com/https-yourwebsite-com-private-credit-firms-eye-ai-after-meta-deal/

Tech Stocks Slide Amid AI Concerns: Market Jitters in 2025