Self-learning control for energy intensive industries, introducing Gigaton

Today, Carbon Re becomes Gigaton.

For years, we have deployed AI on top of legacy control systems in industrial plants… we were moderately successful. We hadn’t yet learned what it took to be wildly successful at scale.

Our journey to Gigaton is the story of those learnings, their culmination into our success today and the product we believe will define the next era of industrial control.

Why now

The Substitution Era; The Age of AI.

About a year ago, Josh, our CEO, sent the team a note titled “There’s an elephant in every control room.” It put into words something we’d felt for a while and set us on a new direction as a company.

Industrial plants have changed more over the past decade than at any point in history. The average cement producer we work with puts more waste energy into their process today than fossil fuels. They have to in order to remain economically competitive and reduce carbon emissions. Perhaps unsurprisingly, a process designed to burn coal reacts very differently when burning sewage sludge or plastics.

But it’s not just the fuels; it’s the materials as well. Cheaper, recycled materials are an enormous cost and carbon driver for our customers.

Changing the two key inputs into the process - fuels and feedstock - requires them to overhaul the plant itself. So, our customers have bolted on new machinery to manage: new burners, bypass systems, laboratories, etc.

As a result, today’s plants are entirely different to what they used to be.

You would assume that as the process changed, so would the software controlling it. It’s in the name after all - process control software.

This was the elephant in every control room. Control systems - expert systems, APCs - weren’t effective. Too often, they were turned off or limited to the point where operators were effectively running the plant manually. Everyone in the control room knew it, and there were very few exceptions to this rule. Expert systems were failing to meet the demands of modern plants.*

*In very specific cases, such as our partnership with Holcim, extremely close collaboration on and integration with a control system that is owned and maintained by the customer led to fantastic results. But for the majority of the industry, propping up and ‘optimising’ the status quo is not effective.

Internally, we say our customers have been hit by two mega trends: The Substitution Era and The Age of AI. This was the effect of the substitution era, an obsession with more efficient production - substituting anything and everything out for cheaper inputs - which had caused enormous upheaval in the industrial world.

At the same time, the Age of AI officially arrived, fundamentally changing the world as we knew it. Industrial manufacturers have been investing in digitalisation for decades. The plants we work with are effectively giant digital data centres, spitting out terabytes of data every year, and for the first time ever, we have technology capable of using that data effectively. But layering AI on top of fixed rules is illogical.

If process control software needed to be rebuilt to meet the demands of The Substitution Era, that became even more pressing with The Age of AI.

Our customers were starting to demand more. The most advanced customers understood that new infrastructure was required to deliver ROI today and, importantly, to build a future in which plants are fully autonomous.

We, a team of AI scientists, process engineers and software geeks who had spent the past five years in control rooms, were perfectly placed to rethink industrial process control and optimisation from first principles.

Solving this is one of the world’s most challenging control and machine learning problems. Non-linear data, continuous chemistry, complex high-heat combustion, long feedback loops, unimaginably large volumes of materials and often very old machinery. It requires an expert cross-disciplinary team of machine learning, control, automation, software and process engineers. And it requires a deep partnership with amazing engineers at customer sites.

What we’ve built

Gigaton is a Self-Learning Control system that sits at the heart of plant operations. It applies AI to both the control and optimisation layers to directly control actuators, explains its actions to the operator, and, most importantly, learns as your plant conditions change.

Over the coming weeks, we’ll be sharing more about our Self-Learning Control system and why we think its the technology for the The Substitution Era and The Age of AI.

Why Gigaton

A gigaton is one billion tons. It's the unit climate scientists use when they talk about the scale of change that matters. It's the unit we've always been oriented around. It's why we started, and it's the measure we use to judge whether we're making a difference.

Gigaton is also a statement of our commitment to this industry.

We understand the weight of the problems you face and the scale at which you need to operate. We are committed to being the partner who attracts the world’s brightest and most dedicated minds to roll up their sleeves on site with you and build the products you deserve.

Gigaton Technologies Ltd

We are Gigaton - a team of engineers, AI scientists and industry specialists building autonomous control systems for the world's energy-intensive industries. Cement, steel, glass: the industries that produce around 20% of global greenhouse gas emissions. The industries that build our world.

We have always been here for the gigatons. Now we have a name to match.