Data-centre electricity, global
415 → 945 TWh
2024 → 2030, IEA Energy & AI (2025)
§01.0 · The constraint
The energy transition is running out of daylight, land and time.
AI data centres are expected to more than double their electricity use by 2030. On the ground, solar is bound by night, weather and the angle of the sun — it needs batteries to be firm and vast land to be large. Nuclear is firm, but takes 12–15 years to permit and build.
In geostationary orbit, 35,786 km up, a solar array sees the Sun more than 99% of the time — the only gaps are brief eclipses around the equinoxes. No night. No clouds. And the satellite hangs fixed over a single receiving site, so the beam never has to hand over.
Johor data-centre capacity
10 MW → 2.7 GW
2021 → 2027 projected, driven by the JS-SEZ
Clean-energy investment, 2025
$2.2 trillion
IEA World Energy Investment 2025
Malaysia renewable capacity target
70% by 2050
National Energy Transition Roadmap (NETR)
Generation over 24 hours
Same sun. Very different day.
Shape only, normalised to peak. Ground curve is a clear-sky day; cloud and monsoon cut it further. Aphelia's 99% capacity factor is a design target in geostationary orbit, where eclipses are limited to short periods around the equinoxes.
§02.0 · The Aphelia SubTile
One part.
Six layers.
Every satellite is built from one repeated 500 mm SubTile. It collects sunlight on its back, converts it to microwaves on our own GaN chips, and radiates it through helical elements on its front. Scroll to take it apart.
- SubTile0.5 × 0.5 mThe base manufacturing unit
- Tile · 25 SubTiles2.5 × 2.5 mDemonstrator and production unit
- Array · 3,136 Tiles140 × 140 mFirst commercial scale
- Commercial design170 × 170 mPlanned PPA configuration
Same part from the first test to the full constellation. Manufacturing scales incrementally, with no redesign at any step.
§03.0 · Wireless power transmission
A link that adapts to the monsoon.
The phased array steers its beam electronically — the travelling colour across the transmitter is the phase gradient doing it. The link runs at 19.7 GHz through clear skies and light rain. When heavy tropical rain arrives over the receiving station, it switches to 12.2 GHz to hold power. Make it rain.
Atmospheric efficiencies are Aphelia link-budget figures for Malaysian conditions. Locked at 19.7 GHz, heavy rain would cut the link to ~4%; switching to 12.2 GHz holds a 43.65% floor. Visual is illustrative and not to scale.
The energy chain — what we build in-house
- 01SunlightSi/perovskite tandem cells, ~30% efficient
- 02DC → RFAphelia GaN MMIC power amplifiers
- 03Phased arrayHelical elements, five-zone tapered beam
- 04GEO → groundAdaptive 19.7 / 12.2 GHz link
- 05RF → DCRectenna Ground Station converts the beam
- 06GridDirect injection at the substation
Why helical elements
−75%
GaN dies needed per tile versus patch antennas, with double the effective range. Helices are circularly polarised by nature, which rides through ionospheric scintillation better than linear polarisation.
Why a tapered beam
5 zones
Transmit power is tapered across five zones of the array to shape a near-Gaussian beam — concentrating energy on the receiving station instead of spilling it around.
Why GEO
35 yrs
Target life, versus 10–20 years typical in low orbit. Near-continuous sunlight removes the battery system, and the array radiates its own heat — eliminating ~2,000 kg of deployable radiators.
§05.0 · Against the alternatives
Firm, clean baseload — without the batteries or the reactor.
For the next few years, Aphelia's real competition for Malaysian and ASEAN power contracts is on the ground: solar farms with batteries, and small modular reactors. Here is how a satellite fleet compares — including what the first unit costs before scale.
Capacity factor
Solar PV + BESS15–17%
Nuclear SMR90–95%
Aphelia SBSP99%†
Levelised cost of energy, US$/kWh — the learning curve
Aphelia · first 140 m unit$0.319
Aphelia · 50-satellite fleet$0.093
Aphelia · 170 m commercial~$0.072
Solar PV + BESS$0.11–0.14
Nuclear SMR$0.09–0.15
| Johor 2035 fit | Solar PV + BESS | Nuclear SMR | Aphelia SBSP |
|---|---|---|---|
| Land footprint | ~10,000 acres | 1,500–3,000 acres | ~800 acres~90% less than a solar farm |
| Initial CAPEX | — | ~US$100M | ~US$35M~65% below SMR |
| Battery storage | 5.12 GWh, replaced every 10–12 yrs | None | None |
| Cooling water | Minimal | 378–606 MLD | Zero |
| Lead time | 5–7 years | 12–15 years | 3–5 years, phased |
| Treaty & licensing | Low | Very high (NPT, IAEA) | Moderate (ITU filing) |
| Scaling | More land, more batteries | New reactor build | Launch more of the same satellite |
† Design target. Aphelia figures are company projections and modelling, not yet validated at flight scale — the central execution risk, which our roadmap is sequenced to retire. Benchmarks for Johor (JREP 2030, JS-SEZ 2035); full assumptions available on request.
§06.0 · Why Malaysia
Built on the equator, under the geostationary belt.
Geostationary satellites sit directly above the equator. Malaysia sits almost directly beneath them — the shortest, steepest path from orbit to ground, over an arc far less crowded than those above North America, Europe and East Asia.
Equatorial advantage
ITU orbital slots are first-come, first-served, and filings rose ~400% in 2020–2025. Equatorial states have argued for preferential GEO access since the 1976 Bogotá Declaration. Early Malaysia-flagged filings are a strategic asset in their own right.
A live demand shock
The Johor–Singapore SEZ took Johor's data-centre capacity from ~10 MW in 2021 to ~1.3 GW in 2024. Most hyperscaler tenants carry 100% renewable mandates — stricter than the national target, and needing firm power.
Policy already in motion
The National Space Policy 2030, the Malaysia Space Exploration 2030 plan, the Malaysian Space Board Act 2022 and MASIC together target 1% of GDP and 5,000 jobs from space by 2030.
Neutral by design
A Malaysian supplier, from a country with a long non-aligned tradition, for the many buyers that want energy security without depending on a single geopolitical bloc's technology.
§07.0 · Markets
Win Malaysia. Expand across ASEAN. Then go global.
The reference market
Firm renewable baseload for Johor's data-centre corridor, island grids and the national utility — sold under 20-year-plus power purchase agreements.
- Tioman Island case: one satellite covers ~48–60% of island baseload and avoids $100M+ of submarine cable
- JS-SEZ hyperscalers with 100% renewable mandates
- 1% of Malaysia's renewable build-out ≈ 300 MW, ~US$300M in contracts
Islands & archipelagos
The Philippines and Indonesia share the Tioman economics at far larger scale: off-grid islands, typhoon exposure that limits ground solar, and expensive cable links.
- ASEAN regional demand ~2,200 TWh
- ASEAN Power Grid lets shared ground infrastructure serve several grids
Sovereign energy
Energy-security buyers in the Gulf, plus institutional partnerships in Japan and Europe — governments diversifying away from imported fuel and single-bloc technology.
- Global renewable market: ~$1.08T (2024) → ~$1.57T (2032)
- Partnership routes via ESRIC (Luxembourg) and Japan's SSPS programme
1.5 MW
firm, 24/7 power per commercial satellite
~81
satellites to supply Malaysia's 1.05 TWh corporate green-power quota
−90%
land footprint vs a large solar farm (10,000 → ~800 acres)
20+ yrs
typical PPA term — the next five years decide who signs them
Aphelia modelling; targets, not guarantees. Market figures from NETR, IEA and Aphelia market analysis.
§08.0 · The field
The physics is proven. The race is now about cost.
Beaming power is not experimental physics, and continuous GEO-to-Earth radio transmission has run over this region for decades. Space solar has stalled on two problems: satellites too large to launch affordably, and links too inefficient to pay off. A small, efficient, mass-produced tile goes after both.
Peter Glaser publishes the solar power satellite concept in Science.
NASA JPL / Raytheon beam 30+ kW over 1.6 km at Goldstone; rectennas convert over 82% to DC.
MEASAT begins continuous GEO-to-Earth transmission over Malaysia — the same path, carrying signal instead of power.
ESA SOLARIS cost–benefit studies find a positive business case for space solar in Europe.
Caltech MAPLE performs the first wireless power transfer in space, and beams detectable power to Earth.
DARPA POWER sets a power-beaming record: 800+ W over 8.6 km.
Venture capital arrives. Star Catcher, Aetherflux and Reflect Orbital raise more than $175M between them.
Aphelia ground and air-to-ground demonstrations; ITU and national filings.
| Player | Approach | End market | Where Aphelia differs |
|---|---|---|---|
| Aetherflux (US) | Laser beaming | Pivoting to in-orbit AI data centres | Aphelia stays focused on ground PPAs |
| Star Catcher (US) | Satellite-to-satellite beaming | In-space power | Different end market |
| Reflect Orbital (US) | Orbital mirrors onto solar farms | Extending ground solar | Complementary, not a substitute for baseload |
| Virtus Solis (US) · Space Solar (UK) | Microwave SBSP | Grid power | Closest peers; Aphelia claims higher antenna efficiency |
| China (Bishan) · JAXA · ESA | State programmes | National roadmaps | Legitimise the category; potential partners |
| Aphelia (MY) | GEO, GaN phased array, adaptive 19.7 / 12.2 GHz | Utility PPAs, Malaysia & ASEAN first | Equatorial, non-aligned, government-integrated |
§09.0 · Roadmap
Four de-risking stages, each proven before the next begins.
-
2026 – 2027 · now
Stage 1 · WPT MVP TRL 6
Ground demo of the GaN MMIC array, beamforming, phase steering and dual-frequency logic. Air-to-ground demo. ITU and national filings.
-
2027 – 2028
Stage 2 · Small-scale SBSP TRL 7
A compact single-tile satellite in GEO validates the full space-to-Earth chain and real capacity factor. First Malaysian PPA or MOU.
-
2028 – 2030
Stage 3 · Pilot deal TRL 8
Customer's choice: a 10 × 10 m space pilot in GEO, or a 5 × 5 m air-to-ground pilot — both on the same ground receiver. Manufacturing scales in parallel.
-
2030 – 2032
Production scale
Commercial satellite production. First ASEAN follow-on market; partnerships in the Gulf, Europe and Japan.
-
2032 – 2033
Stage 4 · Initial PPA TRL 9
First 170 × 170 m revenue-generating satellite, then a ten-satellite commercial constellation on shared ground infrastructure.
§10.0 · Traction
Early, and already in the right rooms.
ESRIC · Luxembourg
Start-up Support Programme
Selected for Batch 5, one of seven startups worldwide — the first from Malaysia
Startup World Cup 2023
Malaysia national champion
Represented Malaysia at the Grand Finale, San Francisco
Payload Pioneers 2024
Razlan Hamdan, honoree
Payload's annual list of rising leaders in space
Ecosystem
MASIC member
Malaysia Space Industry Consortium, launched 2023
Partner · Canada
Spaceium
Letter of intent on wireless power for orbital service stations
Partner · United States
Astrolab
Cooperation agreement on wireless power technology
§11.0 · Team
A systems-engineering company, not a science-fiction one.
Razlan Hamdan
Co-founder · CEO & CTO
Electrical engineer leading Aphelia's mmWave wireless power R&D with universities and regulators. Payload Pioneers 2024 honoree.
Matt Kamil
Co-founder · CMO
Founded one of Malaysia's largest tech communities. Has taken several university technologies to market, building national category leaders.
Lian Ming Goh
Satellite Systems Engineer
MEng Aerospace, Southampton. 5+ years at Astroscale and AAC Clyde Space; mission design for ELSA-M debris removal and the UKSA/ESA xSPANCION constellation.
Pravindeeran Veerasingam
Business Development
Chemical engineer specialising in procurement, contract management and cost optimisation for large projects.
Ryan Bensonella
Technical Advisor
2024 NASA Research Fellow with 10+ years in aerospace. Led the first autonomous in-space welding demo; worked on Artemis Lunar Gateway, Cygnus and ISS modules.
§12.0 · Work with us
Power your grid from orbit.
We're working with utilities, data-centre operators, governments and partners across Malaysia and ASEAN. Talk to us about pilots, power purchase agreements and partnerships.
- PilotsAir-to-ground or space pilot, both on the same fixed-footprint ground receiver
- Power purchase agreementsFirm, 24/7 renewable baseload on 20-year-plus terms
- PartnershipsGovernment, research, manufacturing and launch