The Architecture

The closed-loop architecture.

Every Lumina Nova hub is a single engineered system: renewable generation, long-duration storage, and compute offtake operating as one closed loop. This page explains how the system is designed, why the integration matters, and how the same architecture adapts across workload cycles.

One system, not three.

Most renewable energy projects solve for generation. Most data centres solve for consumption. Very few solve for both inside a single operational envelope. This separation is the source of two persistent industry problems: renewables struggle with dispatch certainty, and digital infrastructure struggles with power availability. Lumina Nova's thesis is that integrating both ends of the energy equation — behind a single facility boundary, with storage as the bridge — produces a more bankable, more predictable, and more scalable asset than either half built alone.

Our facilities are designed around four principles:

Principle 01

Generation sized to compute

Solar capacity is specified to match the facility's 24-hour compute load, not to maximise export. Generation and consumption are matched variables in the design, not independent ones.

Principle 02

Storage sized to bridge the diurnal cycle

Battery capacity is specified to carry the facility through nightfall and low-irradiance weather without external power. Storage is not an adjunct — it is the piece that makes the closed loop possible.

Principle 03

Compute as the committed offtaker

Every electron has a pre-determined consumer on-site. There is no merchant exposure, no grid export, no dependence on third-party dispatch.

Principle 04

Architectural flexibility across workload cycles

The compute envelope is designed to accommodate both current-generation and next-generation workloads. Cooling capacity, power density, and connectivity are specified beyond immediate requirements so that workload mix can shift without facility retrofit.

Figure 01
The Integration Thesis
Conventional Model SOLAR Project 01 stand-alone Grid dispatch STORAGE Project 02 stand-alone Tariff exposure COMPUTE Project 03 stand-alone Power availability THREE EXPOSURES each developer bears independently vs Lumina Nova Model FACILITY ENVELOPE no grid input no external offtake ENERGY FLOW · ONE-WAY · ON-SITE SOLAR Pillar 01 generate STORAGE Pillar 02 bridge COMPUTE Pillar 03 offtake · underwritten offtake revenue ONE EXPOSURE compute offtake, the one we underwrite

Three stand-alone projects each carry external exposures their developers cannot control. Integrated within one facility envelope, those exposures collapse to a single, underwritable offtake risk.

Site Imagery
Kolam Programme — Perak, Malaysia
Aerial imagery to be added
04°29'N · 101°09'E · Gopeng / Kampar

Generate. Store. Compute.

The pillars are introduced briefly on the homepage. Here, each is described at the level of detail an institutional reader requires to assess feasibility.

Each Lumina Nova facility produces three measurable outputs: renewable energy generated and stored on-site, compute hours delivered to contracted workloads, and the capital efficiency of converting one into the other. The three pillars described below — Generate, Store, and Compute — are the physical systems that produce these outputs.

Pillar 01

Generate

Floating Solar Photovoltaic

Lumina Nova deploys utility-scale floating solar photovoltaic systems on underutilised water bodies across Peninsular Malaysia — former tin mining lakes, reservoirs, aquaculture sites, and industrial water basins. Floating deployment is central to our model for four reasons.

Land is preserved for its primary use. Malaysia's land bank is constrained by palm oil, urban development, and protected areas; utility-scale ground-mounted solar competes directly with these uses. Water bodies do not.

Panel efficiency is higher. Natural cooling from the water surface reduces operating temperature, typically improving energy yield by three to eight percent relative to comparable ground-mounted installations.

Evaporation is reduced. Floating arrays shade the water surface, materially reducing evaporation losses — a secondary benefit of value to water authorities and aquaculture operators.

Permitting is faster. Water-body leases are negotiated with single landowners (state, state agency, or private operator), avoiding the multi-party consolidation typical of large ground-mounted projects.

Our standard pod-based modular design allows deployment to be staged across a water body, with commissioning possible on partial capacity while construction continues elsewhere on the site.

Pillar 02

Store

Grid-Scale Battery Energy Storage

Our facilities integrate grid-scale lithium iron phosphate (LFP) battery energy storage systems specified for long-duration, high-cycle industrial use. LFP chemistry is selected over alternatives for four reasons relevant to 25-year infrastructure: superior cycle life under daily deep discharge, thermal stability appropriate for tropical climates, absence of cobalt in the supply chain, and mature manufacturing scale that delivers competitive capital cost.

Storage capacity is sized to the facility's compute load profile. In typical operation, the BESS absorbs excess solar generation during peak irradiance hours and discharges through the evening and overnight period. This operational pattern allows the facility to maintain consistent compute availability across the full 24-hour cycle without drawing from the national grid.

Storage is not specified as a contingency layer. It is specified as the primary dispatch mechanism that makes off-grid compute operation economically viable.

Pillar 03

Compute

AI-Optimised High-Performance Compute

The compute pillar is where Lumina Nova's long-horizon design philosophy is most visible. We architect each facility for workload flexibility from the first day of operation, because the economics of different compute workloads will evolve over a 25-year asset life in ways that cannot be fully predicted today.

In Phase 1, facility compute capacity is deployed across a workload mix that reflects current economic returns. This includes AI training and inference workloads, high-performance computing applications, and protocol-level compute supporting distributed network infrastructure. The mix across these workload categories adjusts in response to market pricing, customer demand, and capacity availability.

In Phase 2, as AI and HPC contracted pricing and hyperscaler demand patterns mature in the Malaysian and regional market, facility capacity is expected to shift toward longer-duration AI and HPC contracted workloads. This shift does not require facility retrofit because the Phase 1 infrastructure is specified to accommodate it.

Zoned Density Zoned compute halls with differentiated power density envelopes, allowing both standard-density and high-density deployments within the same facility.
Hybrid Cooling Hybrid cooling infrastructure supporting air-cooled, immersion-cooled, and liquid-cooled deployments as workload mix evolves.
Fibre Connectivity Fibre connectivity specified and deployed from day one, with diverse routing to regional carrier hotels, eliminating the most common retrofit obstacle.
Power Envelope Power envelope and switchgear sized beyond initial requirements, allowing higher-density deployments without upstream electrical retrofit.

The closed loop, visualised.

A conceptual render of a Lumina Nova facility. Six distributed floating solar clusters, a land-based compound housing storage and compute, and the cable network that ties the system together. The visualisation runs a compressed 24-hour cycle — watch generation ramp through the day, storage charge through midday, and the BESS dispatch overnight to keep compute running on stored power.

Facility Concept
Initialising visualisation
Scroll into view to activate
Facility Time
08:00
Morning · Generation peak
BESS State of Charge
45%
Charging
Energy Flow
Solar Generation
Storage Charge / Dispatch
Compute Offtake
Drag to orbit · Scroll to zoom
Pillar 01
Floating Solar
Pillar 02
BESS Farm
Pillar 03
Compute Hall
Conceptual render · 24-hour operational cycle Expand to fullscreen

Why integration changes the economics.

A floating solar project, built as a stand-alone asset for grid export, is subject to three external variables its developer cannot control: grid connection timeline, feed-in tariff structure, and curtailment risk during oversupply periods. Each of these directly affects revenue.

A data centre, built as a stand-alone asset, is subject to three external variables its operator cannot control: electricity tariff structure, power availability during grid constraints, and tariff exposure during policy revisions.

Integrating both within a single facility converts these external variables into internal design parameters. Revenue is determined by compute offtake pricing, not by grid dispatch decisions. Input cost is determined by facility capital structure, not by retail electricity tariffs. The two revenue exposures are collapsed into one, and the one remaining exposure is the one our operating model is built around.

The result is an asset whose cash flow profile is closer to contracted infrastructure than to merchant power — while retaining the commodity exposure upside of compute demand growth. That combination is rare. It is the reason institutional capital finds the model attractive, and it is the reason we believe it is defensible across market cycles.

Figure 02
From Three Risks to One
Stand-Alone Assets RISK 01 Grid dispatch & tariffs RISK 02 Power availability RISK 03 Curtailment & policy THREE EXPOSURES external · uncontrolled integration within facility envelope Integrated Facility ONE EXPOSURE Compute offtake pricing & demand ONE EXPOSURE the one we underwrite

Three risks a stand-alone developer cannot control become one risk an integrated platform is built to manage. Revenue predictability is the product.

Revenue resilience by architecture.

No infrastructure asset is immune to cycles. Compute workload pricing, energy market dynamics, and protocol-level economic events all produce revenue variability across a 25-year horizon. Lumina Nova's response is to design resilience into the facility, the portfolio, and the capital structure — rather than rely on any single source of stability.

Facility level
Workload flexibility allows compute capacity to shift between revenue categories as relative pricing changes, without physical reconfiguration.
Portfolio level
Projects are sequenced rather than built in parallel. Pilot operations inform subsequent sites. No single facility represents concentration risk to the platform.
Capital structure level
Compute revenue is deployed through a structured treasury framework designed to smooth cash flow volatility across market cycles. The framework integrates Shariah-screened digital asset instruments with stable-value yield strategies, providing a second-layer resilience complementary to the physical redundancy of the facility itself. Governance of this framework is formalised under dedicated Shariah oversight, with ring-fenced policy separating capital-stack-specific revenue streams.

The principle is simple: redundancy should exist at every layer of the platform, not only in the physical plant.

Aligned with Malaysia's transition.

Lumina Nova's platform architecture is designed to contribute to — and benefit from — three Malaysian policy priorities.

Priority 01

National Energy Transition Roadmap

The NETR establishes a national objective of expanding renewable generation capacity without overburdening grid infrastructure. Off-grid, self-consumed renewable generation directly supports this objective by adding renewable capacity in a form that does not require incremental grid investment.

Priority 02

Bank Negara Malaysia Digital Asset Tokenization Roadmap

BNM's tokenization roadmap identifies infrastructure, Islamic finance, and sustainability-linked assets as priority categories for tokenized capital markets development. Lumina Nova's asset class sits at the intersection of all three.

Priority 03

AI Nation 2030 & Digital Economy

Malaysia's broader digital economy positioning — including its regional role in AI and data centre infrastructure — is constrained by power availability. Self-powered digital infrastructure expands national compute capacity without competing for grid resources needed elsewhere.

These alignments are not incidental. The platform was designed around them.

Staged, not parallel.

Lumina Nova's deployment approach reflects a deliberate preference for operational proof over scale acceleration.

The first facility is commissioned to full commercial operation before subsequent facilities advance to construction. Twelve months of operating data from the pilot informs design decisions, capital structure, and operating protocol on follow-on sites. This approach extends the time to full portfolio deployment, but materially reduces correlated execution risk across the platform.

Follow-on facilities incorporate learnings from the operating pilot in three areas: facility engineering specifications, compute workload mix and vendor selection, and operational staffing and support structures. By the time the portfolio reaches its target scale, each facility in sequence benefits from the operating experience of those that came before it.

This is the infrastructure discipline the model requires. We prefer to build it in from the start.

A platform engineered for Malaysia's long horizon. Designed to adapt across every cycle it operates through.

A platform engineered for Malaysia's long horizon. Designed to adapt across every cycle it operates through.

Discuss the Architecture