An emerging artificial intelligence company may be offering one of the clearest signals yet that digital optimisation can deliver meaningful emissions reductions in heavy industry. Deep.Meta, a UK startup founded on physics-driven modelling, has achieved close to a ten percent reduction in carbon emissions at Spartan UK, the country’s sole steel plate manufacturer. The result, validated during early testing in Newcastle upon Tyne, sets the stage for a full live pilot at the plant and establishes the firm as a potential contributor to the sector’s long-term transformation.

The British steel sector remains small in size compared to its global counterparts but plays an outsized role in national manufacturing. Spartan UK is the only domestic producer of plate steel, a material central to construction, infrastructure and defence manufacturing. As the UK pursues increasingly ambitious climate targets, the ability to reduce emissions from processes like reheating, rolling and finishing has become a defining challenge. Globally, steel production accounts for nearly nine percent of total carbon emissions. The industry has historically relied on incremental efficiency upgrades rather than systems-level redesign, leaving many facilities locked into high energy consumption. Investors and governments have begun pressuring mills to demonstrate that cleaner production methods can align with financial performance. In this context, a ten percent reduction in emissions from operational optimisation rather than equipment replacement is a significant milestone.

Read more: Digital Realty Expands Frankfurt Footprint with New AI-Optimised, Renewable-Powered Data Centre

The company’s technology, known as Deep.Optimiser PhyX, combines real-time sensor feeds with material science models to construct a digital twin of furnace operations. This twin predicts temperature changes, tests alternative heating profiles and simulates hundreds of production cycles in a fraction of the time required for traditional adjustments. The platform’s goal is to tighten control over parameters that influence energy usage, such as dwell time, furnace zoning and slab sequencing. By modelling how the steel behaves under different thermal and mechanical conditions, the system identifies opportunities to reduce fuel consumption without disrupting throughput or compromising quality. Founder and chief executive Dr Osas Omoigiade said that achieving net zero is impossible unless steel producers adopt far greater precision in process management. He emphasised that integrating physics directly into the AI training process has been a turning point, allowing the system to make predictions anchored in industrial reality rather than purely statistical patterns. The company sees its progress at Spartan UK as an important proof case with international relevance. Deep.Meta has stated its ambition to prevent ten megatonnes of carbon dioxide emissions by 2030, and the steel plate facility provides a controlled environment to gather evidence for a wider commercial rollout across the UK and abroad.

The manufacturing sector has historically been cautious about adopting machine learning tools due to concerns that opaque models could generate unstable or ill-informed recommendations. Deep.Meta has attempted to address this hesitation by embedding physical laws into its AI structure, ensuring that predictions follow known thermodynamic and metallurgical behaviours. Senior machine learning scientist Dr Kwangkyu Alex Yoo noted that industrial buyers want algorithms that do more than generate predictions. They need systems that behave predictably under pressure, anchor their outputs in established principles and offer explanations that operators can interrogate. By merging learning algorithms with physics-grounded equations, Deep.Meta argues that it has created a new category of operational optimisation that is both trustworthy and scalable. The approach could be particularly attractive to producers facing rising carbon prices and more stringent efficiency mandates.

Since 2020, Deep.Meta has raised more than 2.1 million pounds to advance its modelling architecture and expand its industrial partnerships. The company is now competing in the UK government’s Manchester Prize, an AI-focused innovation challenge that will award one million pounds to the winner in 2026. Funding from the programme is being used to deepen the physics-based integration of the digital twin, improving accuracy in predicting heat transfer, timing and equipment behaviour. Stronger predictive performance could translate directly into further energy savings for partner facilities, including Spartan UK and other mills evaluating the technology.

Explore OneStop ESG Marketplace: AI (Artificial Intelligence)

The steel industry’s decarbonisation journey has often been defined by large, capital-intensive solutions such as hydrogen-based direct reduction and electric arc furnace transitions. While those technologies remain essential for long-term transformation, Deep.Meta’s work underscores that significant emissions reductions are also achievable through digital control and improved thermal efficiency. If the upcoming pilot phase confirms the early results, the technology may offer a pathway for producers who cannot immediately overhaul their equipment but urgently need to cut emissions and energy costs. This potential positions Deep.Meta not only as a promising startup but also as a contributor to the competitive and environmental future of UK steelmaking.

Explore ESG Solutions on our marketplace - OneStop ESG Marketplace.

Keep abreast of the top ESG Events on OneStop ESG Events.

OneStop ESG Educate: Your go-to source for top ESG courses and training programs tailored to your needs.

Stay informed with the latest insights on OneStop ESG News.

Discover meaningful career opportunities on OneStop ESG Jobs.