Bog'lanish

Energy for data centers and AI compute

Energy for critical digital infrastructure.

Aphelium reduces the energy consumption of existing data centers through software for the optimization of cooling and power systems, and plans the energy infrastructure of new AI compute capacity: power, storage and cooling.

Two directions

Existing facilities and new capacity.

01

Existing facilities

Software for the optimization of cooling and power systems. Operates on existing infrastructure, within limits defined by the operator, without changes to equipment.

Solutions

02

New capacity

Energy infrastructure for AI compute: power availability, grid connection, on-site generation, storage and cooling, planned as one system from the site stage.

Energy Infrastructure

The problem

AI compute is limited by power, and every megawatt is paid for again in cooling.

2×

Data center electricity use expected by 2030.

Manba: IEA.

15–30%

Cooling energy reduction reported for AI cooling control in production.

Manba: Published industry results, 2024–2026.

800 MW

New AI compute capacity agreed for Uzbekistan through 2030.

Manba: Government of Uzbekistan, 2026.

STOYKA QATORISOVUQ YO'LAKISSIQ YO'LAKCRAHCHILLER

How it works

From read-only connection to automatic control.

  1. 01 Week 1

    Connect

    Read-only link to the BMS, power meters and temperature sensors. Any vendor, old or new.

  2. 02 Weeks 2–4

    Model

    A physical model of the facility: heat loads, airflow, chiller curves, tariffs.

  3. 03 Month 2

    Recommend

    Setpoint changes, each with the expected kWh saving and the reason. Your team applies them.

  4. 04 Month 3+

    Control

    Automatic control inside hard limits. Operators can override at any time.

Savings estimate

An estimate from four figures.

Tarif

From 1 June 2026, large enterprises Tarif manbasi

12 MW
1,50
70%
1 100 soʻm 1 kVt·soat uchun
Yillik tejash bahosi
6,07 mlrd soʻm – 12,14 mlrd soʻm
Yiliga tejaladigan energiya
5,52 mln – 11,04 mln kWh
PUE oldin / PUE keyin
1,50 → 1,35–1,43
IT energiyaTejash mumkinQolgan qo'shimcha sarf
Bu faqat baho. Haqiqiy tejash faqat o'qish rejimidagi ulanish bosqichida o'lchanadi.Hisoblash formulasi →

Principles

Four principles govern every deployment.

No hardware swap.

Works through the controllers and meters you already have, across vendors and equipment ages.

Explainable decisions.

Optimization over a physical model with hard constraints. Every setpoint comes with its reason.

Limits first.

Temperature, humidity and redundancy limits are set by your team and are never traded for savings.

Performance-based contracts.

Fees are linked to the measured reduction in energy consumption, verified against an agreed baseline.

By role

What Aphelium means for each role.

  • What changes.
  • Why it changes.
  • How to roll it back.

Central Asia

Built where it is hot, mixed and growing.

Hot

Hot summers mean cooling dominates.

Mixed

Mixed equipment from many vendors and decades.

Growing

Tariffs rising and capacity growing fast.

Energy Assessment

Aphelium conducts a read-only assessment of the facility's cooling and power systems and presents the identified savings potential in kWh and financial terms before any change is made to operations.