Brookfield's $1.3B Bet on AI Infrastructure Investing

Brookfield Asset Management is redefining the AI landscape. In mid-2026, its Radiant platform reached a landmark $1.3 billion valuation, signaling a transition toward physical AI infrastructure assets and compute-as-a-service models. This shift represents a broader movement away from purely speculative software bets toward the industrial backbone of the digital economy, where AI infrastructure investing focuses on the hardware, power, and facilities required to sustain modern processing needs.

The Radiant Merger: Decoding the $1.3 Billion Valuation

The evolution of the artificial intelligence market has reached a critical inflection point where the digital and physical worlds converge. At the center of this movement is Brookfield’s Radiant, a platform that recently solidified its market position through a strategic merger with the London-based cloud computing firm Ori Industries. This transaction significantly altered the valuation landscape for infrastructure-heavy AI firms. By early 2026, Brookfield Asset Management's artificial intelligence infrastructure company, Radiant, was valued at $1.3 billion following its merger with the London-based cloud computing firm Ori Industries in early 2026.

This merger was not merely a consolidation of assets; it was a tactical play to master GPU orchestration across a global network. Ori Industries brought a sophisticated software layer capable of managing distributed workloads, while Brookfield provided the deep pockets and physical real estate. One of the most telling financial details of this deal was the clearing of approximately £11.3 million in debt, which prepared the combined entity for a massive scale-up. As a result, Radiant has seen a nearly 20-fold increase in compute mandates within the UK alone, demonstrating how rapidly governments and private enterprises are seeking localized, high-performance processing power.

The rise of what we call sovereign compute capacity is driving these numbers. Nations are no longer content with relying on offshore data centers for their sensitive AI workloads. They want the compute power locally owned and operated. Radiant addresses this by offering a compute-as-a-service model. This allows institutions to rent high-performance chips like Nvidia’s Blackwell series on demand, bypassing the massive upfront capital expenditures typically associated with owning hardware. For many investors, this shift from CAPEX to OPEX for the end-user creates a predictable, recurring revenue stream akin to traditional utility or infrastructure assets.

The impact of national compute capacity policies on private investing cannot be overstated. As governments treat processing power as a national asset, platforms like Radiant become essential utility providers. This allows them to secure long-term contracts that are insulated from the volatility of the retail software market. We are seeing a fundamental change in how the market values these entities—moving from high-growth tech multiples to the stable, yield-driven valuations characteristic of the infrastructure sector.

The Power Bottleneck: Investing in AI Data Center Power Solutions

If compute is the engine of the AI revolution, energy is the fuel. We have observed a dramatic escalation in the energy requirements of the modern data center. A few years ago, a 5-megawatt facility was considered standard. Today, the industry is regularly discussing 50-megawatt installations, with gigawatt-scale projects on the horizon. This massive spike in demand is creating significant energy grid constraints, as existing utility infrastructure was never designed to handle the dense heat and constant draw of AI clusters.

For the modern investor, the key metric is now time-to-power. It is no longer enough to own a data center; you must also own the power generation or have secured access to the grid years in advance. This is why we are seeing increased interest in investing in AI data center power solutions. Companies that can provide on-site power generation, such as small modular reactors or advanced fuel cells, are becoming the most valuable players in the ecosystem. Brookfield’s extensive portfolio in renewable energy provides it with an organic advantage here, allowing it to bypass some of the queueing issues currently plaguing its competitors.

The table below illustrates the rapid evolution of power requirements that investors must navigate:

When assessing the ROI of gigawatt-scale AI infrastructure projects, the calculation must include the cost of cooling and grid fortification. Standard air cooling is often insufficient for the high-performance compute clusters required for large language model training. Liquid cooling and high-density heat exchange systems are now mandatory considerations. Consequently, the value of the asset is increasingly tied to its efficiency—specifically its performance-per-watt. Investors are shifting their focus from simple square footage to the intricate plumbing and wiring that facilitates high-performance computing.

The Four Pillars: Why Brookfield is the Premiere Physical Asset Play

To understand the broader trend of AI infrastructure investing, one must look at how firms like Brookfield are organizing their capital. They are moving beyond the traditional role of a landlord to become a vertically integrated operator. We categorize this strategy into four distinct pillars that define the modern physical asset plays in the AI compute market.

First, there are the factories. Brookfield’s $30 billion joint venture with Intel to build semiconductor fabrication facilities in Arizona is a prime example. This ensures a steady supply of domestic hardware, mitigating geopolitical risks associated with international supply chains. Second is the power pillar. By leveraging a massive fleet of hydropower, wind, and solar assets, they provide the green energy required by hyperscalers who have strict ESG mandates.

The third pillar is compute. This is where Radiant and the Ori Industries merger play their part, offering the orchestration and management of the actual chips. Finally, there are the adjacencies. This includes the mechanical systems, modular data center growth, and specialized cabling required to interconnect thousands of GPUs. Traditional digital infrastructure REITs often struggle with this level of complexity because they are built for generic server housing, not the hyper-specialized frontier data centers that AI demands.

Modular AI data center assets are particularly attractive right now because they allow for rapid deployment in areas where grid capacity is limited. Instead of building a traditional concrete facility over five years, a modular unit can be deployed in months, complete with its own on-site power generation. The growth potential of modular AI data center assets represents a significant acceleration in how quickly infrastructure firms can realize a return on their capital.

Investor Analysis: How to Analyze AI Infrastructure Stocks Like Brookfield

When looking at the equity side of this equation, investors must adapt their analysis to modern realities. Determining how to analyze AI infrastructure stocks like Brookfield requires a departure from traditional tech metrics like user growth or monthly active users. Instead, we look at backlog equity and the scale of the capital program. Brookfield’s $100 billion AI infrastructure program is perhaps the clearest indicator of their long-term conviction.

One of the most important things to monitor is the conversion of these investments into realized revenue through the compute-as-a-service model. Comparing this to private competitors like CoreWeave reveals a market hungry for reliable, physical-backed capacity. Investors should look for performance moats, such as proprietary cooling technologies or exclusive power-purchase agreements, which serve to protect margins against commoditization.

Furthermore, we must watch the transition from software-focused trades to the ownership of the power, memory, and cabling systems. The "picks and shovels" of the AI era are no longer just chips; they are the high-voltage transformers and the fiber-optic interconnects. When you are investing in chip compute capacity, you are effectively investing in the infrastructure of the future grid.

Strategic moves like the Radiant merger illustrate that the winners in the AI race will not just be those who write the best code, but those who own the land, the power lines, and the hardware that makes the code run. For the long-term portfolio, this represents a stabilization of the AI opportunity. It moves the conversation from speculative bubbles to the tangible, cash-flow-producing reality of global infrastructure.

FAQ

What is considered AI infrastructure for investing?

AI infrastructure refers to the physical layer of the artificial intelligence economy. This includes data centers equipped for high-density computing, power generation and transmission facilities, semiconductor fabrication plants (fabs), and the orchestration software that manages GPU clusters. It is the industrial backbone that allows AI models to be trained and deployed.

How can individual investors gain exposure to AI infrastructure?

Investors can gain exposure by targeting asset management firms that specialize in infrastructure, such as Brookfield (BAM), or by looking at digital infrastructure REITs that are pivoting toward high-density AI housing. Additionally, companies involved in the electrical grid, cooling systems, and specialized hardware manufacturing offer indirect but essential exposure to the sector.

What are the key sectors within AI infrastructure?

The industry is generally divided into four segments: power and energy (renewables and grid stabilization), physical real estate (frontier data centers), hardware manufacturing (semiconductors and fabs), and compute orchestration (software-defined infrastructure and compute-as-a-service).

How does AI infrastructure differ from traditional cloud computing?

Traditional cloud computing is focused on storage and general-purpose processing, which can be housed in standard data centers. AI infrastructure requires significantly higher power density, advanced liquid cooling, and high-speed interconnects between chips to handle the massive mathematical workloads required by large language models. The energy draw for AI is often ten times higher per rack than traditional cloud servers.

What is the expected growth rate for the AI infrastructure market?

While specific figures vary by sub-sector, the move toward gigawatt-scale computing suggests a massive expansion. With compute demand doubling every few months, the capital required for the physical backbone is expected to reach trillions of dollars over the next decade as global capacity shifts toward regional and sovereign compute models.

Are AI infrastructure stocks overvalued?

Valuation is moving away from speculative tech multiples toward infrastructure-style metrics. While some hardware companies see high volatility, firms like Brookfield that own the underlying land, power, and long-term contracts are often valued based on their cash flow and strategic moats. The key for investors is to distinguish between companies that "hype" AI and those that possess the physical assets necessary to facilitate it.