EPG Raises $100M+ Series B to Scale AI Data Centers Globally

Energy Park Group (EPG), a Hong Kong-based AI data center developer, has closed over $100 million in Series B financing led by Crescent Point Energy and Kayne Anderson Capital Advisors — a funding round that arrives as hyperscalers and AI companies scramble to secure infrastructure capacity across Asia-Pacific. The capital will fuel EPG's aggressive expansion into markets where power availability, not land, has become the binding constraint on data center growth.

The raise marks a significant vote of confidence in EPG's integrated model: the company doesn't just lease data center space — it develops the power infrastructure underneath it. That matters in regions where grid capacity lags demand by years, and where securing megawatt allocations requires navigating utility relationships, regulatory frameworks, and long-dated power purchase agreements that most pure-play developers can't or won't touch.

Crescent Point Energy, the lead investor, brings more than capital. The firm specializes in energy transition and infrastructure development — domains that increasingly define who wins in the data center arms race. Kayne Anderson, a longtime player in energy and infrastructure investing, adds strategic weight. Both investors are betting that EPG's power-first approach will prove decisive as AI training workloads push data center power density to levels traditional facilities can't support.

"This investment accelerates our mission to deliver sustainable, high-performance infrastructure for AI and cloud computing," said Dominic Choy, CEO of EPG, in a statement. The company didn't disclose specific deployment timelines or project locations, but the funding announcement referenced expansion across "key markets in Asia-Pacific" — a deliberate vagueness that likely reflects ongoing negotiations with utility providers and government agencies in multiple jurisdictions.

The data center industry is undergoing a quiet but fundamental shift. For years, developers competed on real estate metrics: proximity to fiber, land cost, tax incentives. Power was assumed — a utility you plugged into, not a product you built. AI changed that assumption overnight.

Training large language models and running inference workloads at scale requires power densities that can exceed 100 kilowatts per rack — ten times the load of traditional enterprise IT. That means a single AI data center can demand as much electricity as a small city. In markets where baseload power is already constrained, that's not a procurement problem. It's a development problem.

EPG's pitch is that it solves the development problem. The company positions itself as both data center operator and power infrastructure developer, handling everything from substation buildouts to renewable energy integration to grid interconnection approvals. In markets like Hong Kong, Singapore, and emerging Southeast Asian economies, that integrated capability isn't a nice-to-have — it's the only way to secure multi-megawatt allocations on timelines that matter to hyperscale customers.

The financing announcement didn't detail EPG's current portfolio capacity or pipeline, but the company has previously disclosed projects in development across Hong Kong and regional markets. What the announcement did emphasize: the capital will fund both facility construction and "strategic power infrastructure development" — a signal that EPG intends to own, not lease, the energy backbone of its data centers.

Crescent Point Energy isn't a typical data center investor. The firm's portfolio skews toward renewable energy projects, energy storage, and grid infrastructure — assets that generate cash flow from long-term power offtake agreements rather than real estate appreciation. That background matters here. Crescent Point likely sees EPG not as a landlord play, but as an infrastructure play — one where the underlying asset is contracted power capacity, not square footage.

Kayne Anderson, meanwhile, has been active in energy and infrastructure since the 1980s, managing over $50 billion across strategies that include midstream energy, renewables, and real assets. The firm's involvement suggests EPG's model has passed institutional due diligence on both the infrastructure and real estate dimensions — a bar that's higher than it sounds, given how few data center developers have credible in-house power development capabilities.

The timing of the raise is no accident. Capital is flooding into AI infrastructure, but it's concentrating in markets where power is abundant and permitting is predictable — the U.S. sunbelt, Scandinavia, parts of the Middle East. Asia-Pacific has been slower to attract investment, despite demand from regional hyperscalers and AI labs, because power infrastructure remains fragmented and grid access is harder to secure. EPG's thesis is that this supply-demand mismatch creates opportunity for developers willing to do the hard infrastructure work.

Neither investor disclosed their exact stake or valuation, but the "over $100 million" figure suggests this is a substantial equity round — not just a credit facility or project-level financing. That structure matters. It implies the investors are backing the platform, not just individual assets, and expect EPG to deploy capital across multiple projects and geographies rather than optimizing a single site.

EPG's announcement outlined three deployment priorities: expanding AI-ready data center capacity, accelerating power infrastructure development, and entering new Asia-Pacific markets. Translation: the company plans to replicate its Hong Kong model — building both the facility and the power behind it — in markets where AI demand is growing faster than supply.

Asia-Pacific data center investment hit $6.4 billion in 2023, according to CBRE, but the region still accounts for less than 20% of global hyperscale capacity despite representing over half the world's internet users. The bottleneck isn't demand. It's supply — specifically, the mismatch between where data needs to be processed and where power is available to process it.

Singapore, the region's traditional data center hub, has been under a government-imposed moratorium on new facility approvals since 2019 due to power grid constraints. That moratorium has loosened slightly, but capacity remains rationed and expensive. Hong Kong faces similar dynamics: land is scarce, power is controlled by two regulated utilities, and securing multi-megawatt allocations requires years of planning and regulatory coordination.

That's pushed developers into emerging markets — Malaysia, Indonesia, Thailand, Vietnam — where land and power are more available but infrastructure is less mature. EPG's challenge, and opportunity, is that these markets require exactly the kind of ground-up power development the company claims to specialize in. You can't just lease grid capacity in Jakarta or Kuala Lumpur the way you might in Northern Virginia. You have to build substations, negotiate offtake agreements, and sometimes co-develop generation assets alongside the data center itself.

If EPG can execute that playbook across multiple markets, the Series B capital makes sense. If it can't — if power development timelines stretch or regulatory approvals stall — the funding buys time but doesn't solve the structural problem. The company's ability to deliver contracted capacity, not just announced projects, will determine whether this raise was well-timed or premature.

One notable gap in the announcement: no mention of anchor customers or pre-leased capacity. That's not unusual for growth-stage developers, but it does mean EPG is raising capital on the promise of demand, not the certainty of signed contracts. In a market where hyperscalers are announcing massive AI infrastructure commitments, that's a calculated bet — but still a bet.

EPG isn't the only developer trying to crack the Asia-Pacific AI infrastructure market. Established players like Equinix, Digital Realty, and GDS Holdings have deeper pockets and longer track records. What EPG has — in theory — is agility and focus. The hyperscalers move slowly in new markets. EPG is betting it can move faster, especially in jurisdictions where relationships and local expertise matter more than brand recognition.

The real competition may not be other developers at all. It's the clock. AI infrastructure demand is peaking now, but data center construction timelines run 18-36 months from site acquisition to commercial operation — longer if you're building power infrastructure from scratch. If EPG can't deliver capacity before hyperscalers build their own solutions or lock in capacity elsewhere, the window closes.

The Series B announcement leaves several critical details unresolved — details that will determine whether EPG becomes a regional infrastructure leader or a well-funded also-ran.

First: Where exactly is the capital going? "Key markets in Asia-Pacific" could mean anything. Malaysia and Indonesia offer cheap power but complex regulatory environments. Taiwan and South Korea have advanced infrastructure but limited available capacity. The specific markets EPG targets will shape both its risk profile and its competitive positioning.

Second: What's the power sourcing strategy? EPG emphasizes sustainability in its messaging, but the announcement doesn't specify whether the company is building renewable generation assets, buying renewable energy credits, or simply connecting to existing grids with a mix of baseload and renewables. That matters for both climate commitments and operating economics — renewable power is cheaper long-term but requires more upfront capital.

Third: Who are the customers? The company references "AI and cloud computing" demand, but hasn't named anchor tenants or disclosed pre-lease activity. Hyperscalers typically want to see committed capacity before signing long-term contracts. EPG may be in active negotiations, but until contracts are signed, the demand thesis remains speculative.

Read between the lines of the investor roster, and a thesis emerges. Crescent Point and Kayne Anderson aren't backing a landlord. They're backing an infrastructure developer that happens to operate in the data center sector. The bet isn't on real estate appreciation or occupancy rates. It's on EPG's ability to secure and monetize power capacity in markets where power is the scarce resource.

That's a different risk-return profile than traditional data center investing. If EPG executes, the returns could be outsized — infrastructure assets with long-term contracted revenue and embedded power advantages trade at premium multiples. If execution stalls, the capital goes into stranded development costs that don't generate cash flow. There's no middle outcome where EPG becomes a generic data center landlord. The model is too capital-intensive and the power infrastructure commitments too illiquid.

The EPG Series B is part of a larger recalibration happening across data center financing. Traditional real estate investors — REITs, pension funds, family offices — have historically dominated the sector, drawn by stable cash flows and long-term lease contracts. But AI infrastructure requires different capital. It's lumpier, riskier, and more reliant on power infrastructure development than traditional enterprise IT hosting.

That's why energy infrastructure investors like Crescent Point and Kayne Anderson are showing up. They understand contracted power offtake. They're comfortable with multi-year development timelines. And they're willing to underwrite assets where the primary cash flow driver is electricity sales, not rent per square foot. EPG's ability to attract this type of capital suggests the company has successfully repositioned itself from "data center developer" to "power infrastructure platform with data center customers."

EPG enters a crowded but fragmented competitive landscape. Below is a snapshot of how the company's positioning compares to established players across key dimensions.

EPG's strategic bet is clear: go where the incumbents aren't, build what they can't, and move faster than they will. That works if execution is flawless. If not, the company risks getting stuck in secondary markets with stranded assets and no path to premium exits.

The Series B capital gives EPG runway — probably 18-24 months, depending on burn rate and deployment pace. That's enough time to prove the model in one or two markets. It's not enough time to become a regional platform without follow-on capital. Which means the next 12 months will likely define whether EPG raises a Series C from infrastructure funds and becomes a scaled platform, or gets acquired by a larger player looking to enter Asia-Pacific with a built-out power strategy.

EPG's success won't be measured by press releases. It'll be measured by megawatts online, contracts signed, and cash flow generated. Here's what external observers should track over the next 12-18 months:

First, market entry announcements. If EPG is deploying $100 million+ across Asia-Pacific, we should see specific project announcements in new markets within six months. Vague "expansion plans" won't cut it — investors and customers need to see shovels in the ground.

Second, anchor customer contracts. The company needs to announce at least one hyperscale or AI-focused anchor tenant within the next year. Without named customers, the demand thesis remains theoretical — and the capital becomes a bet on speculation, not execution.

Third, power infrastructure milestones. EPG's differentiation is its integrated power model. That means we should see announcements of substation commissioning, renewable energy offtake agreements, or grid interconnection approvals — tangible proof that the company is doing what it says it can do.

Fourth, follow-on financing or M&A activity. If EPG executes well, expect either a Series C from infrastructure-focused funds or acquisition interest from larger data center operators looking to buy rather than build their Asia-Pacific presence. If execution stalls, expect silence — or a down-round.

Strip away the corporate language, and EPG's Series B is a bet on a simple thesis: AI infrastructure will be built wherever power is abundant, not wherever data happens to live today. That's a structural shift with massive implications for where capital flows, which markets attract investment, and who wins the next decade of data center development.

If EPG is right — if Asia-Pacific becomes a major AI infrastructure hub despite grid constraints, and if power-first developers outcompete traditional landlords — this Series B will look prescient. If the thesis doesn't hold — if hyperscalers build their own infrastructure or concentrate capacity in markets with mature grids — EPG becomes a cautionary tale about capital chasing trends before infrastructure catches up.

The next 18 months will tell us which story we're watching. For now, EPG has capital, credible backers, and a clear thesis. What it doesn't have yet is proof. That's what the $100 million is for.

The clock starts now.