MWC 2026: How Huawei’s AI is Revolutionising Construction
At Mobile World Congress (MWC) 2026, Huawei delivered a clear message: the intelligent transformation of traditional industries, from cement to chemicals, has crossed into a new phase.
In light of this, Construction Digital sat down with Charles Li, President of Huawei’s Chemical & Building Materials Business Unit (BU), to explore how AI is shifting from an assistant to a core production driver – ushering in a new era of efficiency, safety and sustainability.
“I believe that 2026 is not only the deep-water zone of transformation but also a singularity moment for industry development,” Charles says.
“In recent years, most AI applications have been concentrated in auxiliary production scenarios, such as detecting whether employees are wearing safety helmets. The biggest change now is that AI is moving from a supporting role to a leading role, truly penetrating core production systems.”
That shift is already catalysing measurable progress. At Conch Cement, Huawei and partners are deploying an optimisation solution that integrates industry mechanisms with AI foundational models.
“By analysing more than 100 characteristic parameters – such as raw materials and processes – in real time, it precisely controls the free calcium range,” Charles adds.
“This has reduced coal consumption by 1%, saving about US$250,000 annually per production line.”
Infrastructure with purpose, not hype
For many manufacturers, high upfront costs and long construction cycles for digital infrastructure remain a deterrent for adopting technologies like AI.
However, Charles is candid about why legacy approaches fall short.
He says: “Traditional IT construction often follows a build first, use later approach – paving the road before driving the car. However, this is difficult to sustain in traditional industries characterised by heavy assets and low profit margins.
“We advocate a new model called driving construction through application.
“Simply put, do not digitalise for the sake of digitalisation – instead, working towards the infrastructure construction around solving specific production and operational challenges, namely AI value scenarios.”
Instead, Huawei starts from specific value scenarios.
“For instance, many manufacturing enterprises face high energy consumption. We introduce AI models specifically for that scenario and then iteratively upgrade the network, data acquisition and computing power foundations on demand.
“Through this approach, every cent invested yields tangible results in improving quality and enhancing safety.”
Budget constraints do not have to be a blocker, either.
“If an enterprise has limited investment for digital infrastructure, it can adopt a public cloud model, renting services based on usage,” he continues.
“This shifts the paradigm from a one-time investment to a monthly pay-as-you-go model, turning CAPEX into OPEX and lowering the threshold for intelligent transformation.”
How AI enables real-time optimisation
Charles is keen to stress that AI is already delivering hard commercial value in core processes – going beyond just pilots and experiments. The Conch Cement case is one example.
“Another example is our practice at Yuntianhua,” he details. “In the chemical industry, the gasifier is like a human heart. Previously, controlling it relied entirely on the experience of veteran experts, so adjustments often suffered from latency.
“We adopted an integrated mechanism + AI approach to build a Real-Time Optimization (RTO) foundational model for the gasifier. It can sense the coupling relationships of hundreds of parameters in real time and automatically find the optimal balance point.”
The outcomes are impressive: each furnace now saves more than US$1.5m in annual costs, achieves a 1.33% reduction in specific coal consumption and boosts the automation rate to exceed 95%.
“This means that AI not only makes production more stable but also directly translates into profit,” Charles applauds.
Providing ‘smart eyes’ for production lines
At Huawei’s MWC booth, one solution in particular caught the attention of visitors: AI-powered rubber quality inspection.
“Visual inspection is common, but defects in chemical products are often microscopic and vary in shape,” Charles says, highlighting the importance of this focus area.
“In nitrile rubber manufacturing, the traditional manual inspection environment is harsh, human eyes are prone to fatigue and the missed detection rate is high.”
Huawei’s system is based on an AI large vision model.
Unlike conventional small models, large-scale models deliver far greater generalisation power.
At PetroChina Lanzhou Petrochemical, the system equipped the production line with intelligent smart eyes, lifting quality inspection accuracy from 70% to 95% and enhancing overall efficiency by more than 15%.
Charles says: “More importantly, this does not merely replace the human eye – it realises fully digitalised traceability of quality data across the entire process and liberates workers from arduous labour.
“This represents the true warmth of technology.”
Safety first
Beyond efficiency and cost, Charles insists that safety is the non-negotiable baseline.
“Safety is our top priority,” Charles asserts. “We have introduced AI into predictive maintenance and hazard identification.”
For instance, when monitoring hazardous chemical gas leaks, conventional point sensors often leave blind spots.
By integrating spectral remote sensing imaging with AI, Huawei can detect leaks within seconds and pinpoint their exact locations.
Similarly, for large rotating equipment, Huawei uses the Pangu predictive model alongside vibration and temperature data to move from a break-fix approach to predictive maintenance.
In collaboration with Changqing Oilfield, this method has boosted the identification accuracy of hidden risks in special operations to more than 94%.
Charles says: “AI enables us to transition from passive response to proactive prevention, which is revolutionary for chemical safety.”
Breaking down data silos
Industrial data is often fragmented across legacy systems, plants and departments.
Huawei’s strategy focuses on stitching this together for AI.
For Charles, the approach is – and has to be – “network first, unified foundation”.
“We use advanced network technologies to connect previously scattered production, office and R&D data,” he clarifies. “Meanwhile, we provide an architecture based on cloud-edge collaboration. At the edge, we enable legacy devices to speak, collecting real-time data. In the cloud, we govern this data through a unified data lake and platform.
“Only when data flows can AI foundational models have the fuel to truly unleash their value.”
Putting safe, transparent AI in control
With AI entering core control loops, concerns about misjudgements and hallucinations in critical systems are obviously going to crop up.
But for Charles, he is keen to emphasise the lack of room for error.
“Our technical path is mechanism + AI,” Charles explains. “We do not hand over control completely to a black-box AI – instead, we integrate physical and chemical mechanism models, accumulated by the industry over decades, with AI foundational models.”
This mechanism model, Charles details, serves as a defined safety boundary, within which the AI operates to identify the optimal solution without exceeding those limits.
Making scale real and human
Many in the industry are sceptical about scaling AI beyond pilots.
On this point, Charles argues that this barrier is finally being broken.
“This is a common pain point across the industry, but we have already seen breakthroughs in the chemical sector,” he says.
“The key to large-scale replication lies in the model’s generalisation capabilities and a standardised architecture.”
He adds that customising a model for every single project was previously expensive, but now – thanks to Huawei’s foundational models, which use a pre-training + fine-tuning paradigm – only minor fine-tuning with a small amount of data is required.
He points to refining and rubber inspection as proof points.
“Data can prove this: in the refining industry, our AI optimisation solution for atmospheric and vacuum distillation units has been successfully implemented and has the potential to be replicated across more than 200 similar units in China,” Charles notes.
“In rubber quality inspection, the solution is also scheduled for rollout across hundreds of production lines. When technical indicators translate into definitive ROI, replication is no longer a slogan but a spontaneous choice made by enterprises.”
When asked why chemical and building materials enterprises should choose Huawei over familiar local suppliers, Charles’ answer focuses on depth, not just breadth.
“At Huawei, our advantage lies in our determination to take root downwards,” he says.
“The Chemical & Building Materials BU shortens the distance between technology and business.
“Our engineers wear work uniforms and dive deep into the production frontlines, eating and living alongside customers in cement and chemical plants to jointly refine solutions.”
Partnerships are central to this model.
He concludes: “More importantly, we adhere to the strategy of being integrated, building ecosystems with global partners, including local suppliers.
“Huawei provides the digital shovel – a foundation that supports the shared success of industry partners.”
