Energy Optimization in Germany: AI-Driven Heating for Existing Buildings

Energy optimization in Germany's existing building stock comes down to one thing above all: using artificial intelligence to continuously adjust heating systems to weather, usage and electricity prices, without opening a single wall. Systems such as KUGU EOS guarantee at least 12 percent energy savings and show an average of more than 20 percent across energy, cost and CO₂ emissions.

For housing companies, property managers and energy service providers, this matters because the existing building stock remains Germany's largest untapped efficiency lever in climate protection in 2026. Digital energy management software is still at an early stage across the housing industry, even as legal requirements for evidence and operational management keep growing.

Four questions decide whether investing in automated energy optimization actually pays off for a portfolio.

What automated energy optimization actually delivers in existing buildings

Automated energy optimization in existing buildings means an AI system continuously adjusts the heating system to actual demand: it evaluates weather forecasts, usage data and system condition, then recalculates new setpoints for flow temperature and the heating curve at short intervals. The control logic follows each building's individual behavior across the entire heating season.

Four building blocks form the technical foundation:

A pilot project run by the German Energy Agency (dena) at Fraunhofer IEE's office building in Kassel showed how reliably market-serving and grid-serving control can work in practice. An AI agent there charged electric vehicles specifically during periods of low prices and high PV generation, communicating continuously with the building management system to do so.

How much savings are realistic: a 12 percent guarantee, more than 20 percent on average

For automated heating optimization in existing buildings, a realistic range is a guaranteed minimum saving of 12 percent and an average of more than 20 percent across energy, cost and CO₂ emissions, the figures KUGU reports for its energy optimization system, EOS. These figures hold regardless of manufacturer, regardless of system age, and without any structural work on the heating system or building.

A guaranteed minimum saving of 12 percent makes automated operational optimization one of the fastest economic levers available in existing buildings, because it works without any upfront investment in new plant technology.

Just how robust these figures are shows up in the ten-month pilot phase KUGU ran together with Gewobag across ten properties in Berlin. Between October 2024 and March 2025, energy consumption and costs fell in all ten buildings by at least the guaranteed margin, and so did CO₂ emissions: roughly 260,000 kilowatt-hours of energy and more than €18,000 were saved, alongside over 50 tonnes less CO₂ output. The 12 percent savings guarantee was met or exceeded in every one of the ten properties.

Good to know: A later, more detailed evaluation of the same ten Gewobag properties, covering 1,055 heating days, arrives at higher figures: 313,039 kilowatt-hours of energy and €21,915, at an average saving of 23 percent. The difference comes down to the evaluation period and methodology; both figures are officially confirmed by KUGU and Gewobag. A second case study from Leipzig showed even higher savings within two months, 39 and 33 percent respectively. That's a seasonal outlier from the transitional period at the end of the heating season, not a year-round average.

For decision-makers, this means solid savings can already be proven within the first heating season, without replacing a single boiler.

Why the existing building stock is under digitalization pressure in 2026

Germany's building sector has once again missed its statutory climate target: in 2025, it caused 103.4 million tonnes of CO₂ equivalent, 3.4 percent more than the previous year. According to the Federal Environment Agency, the increase is mostly weather-related: a colder heating season pushed consumption up noticeably.

At the same time, buildings account for around 35 percent of Germany's final energy consumption, as highlighted by the German Energy Agency (dena). Digital energy management and heating control software is still far from widespread in the housing industry.

According to a survey by industry association GdW of more than 1,000 housing companies, only 8.4 percent currently use such software. A further 8.4 percent have outsourced the task to a metering service provider, and 21.2 percent plan to introduce it in the coming years.

What housing companies should look for when choosing a provider

Four criteria matter most when choosing a provider: data access, scalability across the portfolio, verifiability of savings, and integration with existing systems. Checking these four points before signing a contract avoids the costliest mistakes in the fragmented digital energy management market.

Some context: A supplementary GdW analysis of the software landscape in the housing industry shows that no single provider of digital energy management holds a dominant market position. Most companies rely on small, specialized solutions. Comparing providers against these four criteria therefore gives a far more reliable basis than a simple name comparison.

How KUGU EOS uses a digital building twin for optimization

KUGU EOS, KUGU's energy optimization system, builds automated heating optimization on an individual digital building twin. The system first captures core data such as year of construction, renovation status and plant technology, then a three- to six-day live measurement of heating load and temperatures calibrates the twin under real operating conditions.

On the term: A digital building twin is a computational model of a specific building that replicates its thermal behavior, inertia and plant technology. On this basis, the heating output needed for the coming hours can be calculated in advance.

The building twin as the daily computational basis

Once calibrated, EOS calculates the required heating output every day using more than 30 weather parameters and feeds new setpoints into the heating system every 15 minutes. KUGU VIS, the visual information system, then makes the effect of this control visible to property managers through monitoring and operational transparency, for example when a system deviates from the calculated setpoints.

How a digital building twin contributes to demand-based control of an existing multi-family building shows up most clearly in buildings whose plant technology has run unchanged for decades.

Sector coupling through EOS Strompreisdynamik

With the EOS Strompreisdynamik mode, KUGU also links ongoing heating operation to forecast wholesale electricity prices, for instance from the EPEX Spot exchange, connecting the heat side (heat demand, heat pumps and buffer storage) with the electricity side from photovoltaics and battery storage into one economically managed system. How strongly energy efficiency or electricity cost gets weighted can be adjusted according to the portfolio's goals.

Regulation is giving this approach a tailwind: since January 1, 2025, all electricity suppliers have been required under Section 41a of the Energy Industry Act (EnWG) to offer end consumers with a smart meter at least one dynamic electricity tariff; previously, this obligation applied only to suppliers with more than 100,000 customers. That creates the tariff foundation that lets systems such as EOS Strompreisdynamik actually use variable wholesale prices in heating control at all.

Using electricity prices strategically for heating is becoming more relevant still, since heat pumps overtook gas boilers in heating sales for the first time in 2025, with a market share of almost 50 percent, according to analysis from Agora Energiewende.

How the rollout of energy optimization works in existing buildings

Getting started in existing buildings typically works through a manageable pilot group, often five to ten properties, where calibration and results can be checked quickly before the whole portfolio follows. After capturing the core data and running the multi-day live measurement, automated optimization is usually active within a few weeks.

Getting started can also be backed financially: BAFA subsidizes digital systems for energy-related operational and consumption optimization as well as measurement, control and regulation technology under the BEG individual measures scheme, with a base subsidy of 15 percent of eligible costs, without requiring a new heat generator at all. Ongoing costs for remote monitoring and control are also generally recoverable from tenants under Section 7(2) of the Heating Costs Ordinance (Heizkostenverordnung).

Existing practice speaks to how well this scales across larger portfolios: KUGU says it already manages more than 3,500 heating systems and over one million devices and sensors across more than 300,000 residential units in Germany and Austria. Interested housing companies typically first review the approach in a no-obligation demo using their own portfolio data.

Energy optimization becomes a mandatory exercise with real return in 2026

The real lever comes from a timing mismatch: the building sector is again missing its climate target for weather-related reasons, while only around 8 percent of housing companies use digital control software, even though savings of over 20 percent in ongoing operation are already achievable.

Statutory obligations around dynamic electricity tariffs, growing evidence requirements and a 15 percent BAFA subsidy are all shifting the economics further toward automated operational optimization. Evaluating a portfolio by how verifiable, scalable and manufacturer-independent a solution actually is leads to a more robust decision than a pure price comparison.

The pragmatic next step remains a small, clearly defined pilot group of existing buildings, where calibration, savings and data access can be checked within a single heating season before deciding on a rollout across the entire portfolio.

Frequently asked questions about energy optimization in existing buildings

How long does it take before automated energy optimization takes effect in an existing building?

Usually just a few weeks. After capturing the core data, a three- to six-day live measurement of heating load and temperatures follows, then the AI calibrates the digital building twin, and automated operation starts with new setpoints every 15 minutes.

Does the optimization affect living comfort in occupied buildings?

No, the control system adjusts flow temperature and the heating curve to actual demand without falling below target room temperatures. Because no structural work is required, the switch runs in occupied buildings without any intervention in apartments or plant technology.

What costs do housing companies face, and can they be passed on?

Ongoing costs for remote monitoring, control and data analysis of the heating system are generally recoverable from tenants under Section 7(2) of the Heating Costs Ordinance. In addition, BAFA subsidizes digital systems for operational optimization under the BEG individual measures scheme at 15 percent of eligible costs.

Does the optimization work regardless of the heating system manufacturer?

Yes, the AI-based control works independently of manufacturer and regardless of the system's model or age. It connects to the existing control technology and adjusts setpoints without requiring any components of the existing heating system to be replaced.

Who has access to the recorded energy and operational data?

The housing company remains the owner of its own consumption and system data and should secure export rights contractually. Precisely because the provider market is so fragmented, regulated data access counts as one of the most important criteria to check before signing a contract.