The deal
In Edition 2, we asked whether Amazon's storage-backed PPA strategy would become a moat — or open a lane for specialists. This edition answers part of that question. Because while Amazon was signing PPAs in Australia, something structurally bigger was shifting in the market.
According to BNEF's LCOE (Levelized Cost of Electricity) 2026 report, BESS project costs hit record lows in 2025 — even as solar, wind, and offshore wind all got more expensive. The benchmark cost of a four-hour BESS system fell 27% year-on-year to US$78/MWh. Global average turnkey BESS costs now sit in the ~US$110–120/kWh range depending on duration and system configuration. Meanwhile, solar and wind PPA prices rose approximately 9% in North America in 2025, driven by supply chain constraints, policy uncertainty, and surging datacenter demand. (Sources: BNEF LCOE 2026; BNEF ESS Cost Survey 2025; LevelTen Energy Q4 2025)
The gap between falling storage costs and rising plain renewable PPA prices is not a coincidence. It is a structural signal — and the largest buyers in the world are reading it differently from one another.
The engineering read
Plain renewable PPAs produce power on a pay-as-produced basis. The grid sees the output; the buyer settles financially. That worked when datacenters ran predictable workloads on standard CPU racks.
AI infrastructure changes the calculus. You need electrons at specific times, in specific quantities, at specific locations — not just annual renewable matching on paper.
This is why procurement is shifting from virtual PPAs and RECs (Renewable Energy Certificates) toward physical delivery, 24/7 carbon-free hourly matching, and co-located generation. A virtual PPA settles financially against a wholesale price index — it doesn't move electrons to your facility. A storage-backed physical structure improves temporal load matching and delivery certainty. The difference matters enormously when your continuous load is 100–300MW. (Sources: BNEF; Greening Group, April 2026; WBCSD PPA Structures Report)
The capital allocation read
Amazon, Microsoft, Google, and Meta are each solving the same energy problem with completely different capital strategies.
| Hyperscaler | Strategy | Scale | Signal |
|---|---|---|---|
| Amazon | Storage PPAs + nuclear + SMR development | Multi-GW procurement, 2025 | Reliability through diversification |
| Microsoft | Mega-PPA + developer partnership | 10.5GW with Brookfield; ~40GW total | Speed over control |
| Reportedly expanded upstream (~$4.75B) [REPORTED] | Multiple GW co-located energy + compute | Directionally toward ownership of energy reliability | |
| Meta + Amazon | Renewables + nuclear pivot | Multi-GW nuclear + SMR partnerships | Baseload-first strategy |
Google's move is the most structurally significant. Google has reportedly expanded upstream through a major investment involving Intersect Power (~$4.75B), signalling a move beyond traditional PPA structures toward co-located energy and compute infrastructure. Intersect develops energy parks alongside datacenters — bypassing grid interconnection delays averaging 5+ years, with some high-demand regions reporting up to 12 years. Not full vertical integration — but directionally moving toward ownership of energy reliability rather than offtake of it. (Sources: Alphabet IR, Dec 2025; PV Tech; Energy-Storage.news; Lawrence Berkeley National Lab, 2025) [REPORTED]
Microsoft prioritises speed — a 10.5GW deal with Brookfield between 2026 and 2030, aligned with its round-the-clock zero-carbon target. Both Meta and Amazon are incorporating nuclear into their energy mix, including multi-gigawatt agreements and early-stage SMR partnerships — a baseload pivot that was almost unthinkable in corporate clean energy markets three years ago. (Sources: S&P Global; BNEF, Feb 2026; Meta DEF 14A, 2025)
Energy strategy directly determines two of the three variables. Get power wrong — in availability, cost, or reliability — and effective capacity collapses regardless of IT load or cooling infrastructure.
| Setup | Capex | Firming quality | Utilisation | IRR impact |
|---|---|---|---|---|
| Plain PPA | Low | Low | Volatile | ↓ |
| Grid + spot market | Medium | Medium | Constrained | → |
| Renewable + BESS | Higher | High | More stable | Potential ↑ |
Falling BESS costs combined with rising PPA prices shift the economics toward firmed capacity — higher upfront capex, but more stable long-term cash flows, stronger utilisation, and a path to potential IRR uplift.
The IRR (Internal Rate of Return) uplift assumption holds where PPA tenor aligns with debt maturity and capacity utilisation exceeds the breakeven threshold — both require careful structuring at the project finance level. The underwriting question: is that firming premium being priced correctly in today's market?
What this means for the broader market
The BNEF data reveals a tension worth sitting with: BESS costs are falling fast, but plain renewable PPA prices are rising. The economics increasingly favour firmed and co-located structures in markets where reliability and temporal matching command premiums. Developers that can offer firmed, dispatchable supply are commanding those premiums — and winning the best counterparties.
For colocation operators in APAC, the hyperscaler playbook is bifurcating — those with large balance sheets acquiring or partnering upstream; those optimising for speed signing multi-gigawatt PPAs with tier-one partners. Neither path is accessible to a mid-size colocation operator without an aggregator in the middle.
That aggregator role — contracting firmed renewable capacity at scale and reselling it to operators who cannot do so directly — remains the open lane from Edition 2. Falling BESS costs make the economics more viable. Rising PPA prices make the value proposition more urgent.
If the specialist lane exists — and the economics are increasingly suggesting it does — who is actually building it? And what does the capital structure of a clean energy aggregator targeting APAC colocation operators actually look like?
That is the question Edition 4 picks up — using fresh market signals from the ground in Southeast Asia.
| Term | Full name | Plain English |
|---|---|---|
| PPA | Power Purchase Agreement | Long-term contract to buy electricity from a generator at a pre-agreed price |
| BESS | Battery Energy Storage System | Large-scale battery that stores and dispatches electricity on demand |
| LCOE | Levelized Cost of Electricity | Average cost per unit of energy produced over a system's lifetime |
| LCOS | Levelized Cost of Storage | Equivalent metric for storage systems, accounting for charge/discharge cycles |
| REC | Renewable Energy Certificate | Tradeable certificate representing 1MWh of renewable electricity generated |
| SMR | Small Modular Reactor | Nuclear reactor typically under 300MW; more flexible siting than conventional nuclear |
| IRR | Internal Rate of Return | Annualised return metric used in infrastructure investment analysis |
| APAC | Asia-Pacific | Regional shorthand for Asia-Pacific markets |
| BNEF | BloombergNEF | Bloomberg's energy transition research and analysis unit |
| WBCSD | World Business Council for Sustainable Development | CEO-led organisation advancing sustainable business practices |
| [REPORTED] | Verification flag | Sourced from reported industry sources; not independently verified against primary issuer communications |