Some would argue the infrastructure asset class has been slow to join the digital party, with little incentive to disrupt legacy working practices, and high barriers to entry discouraging innovation. But today’s asset managers cannot afford to slip behind the competition.
Rising inflation and market volatility, combined with the pressures of ESG, are all driving the industry towards greater digitalisation. The pandemic has no doubt accelerated global dependence on all things digital: lockdowns have ushered in a new era of hybrid working, while automation is gaining traction across a variety of infrastructure subsectors.
“We had the pandemic over the past two years, now we have a war going on in Ukraine and the worst inflation in a generation,” says Alex Gunz, fund manager at London-based Heptagon Capital. “Every single business is having to think about how they manage costs and build greater preparedness for uncertainty. The conclusion is that digital transformation matters.”
Cutting costs and streamlining operations are at the forefront of infrastructure asset managers’ minds, regardless of sector. Firms are turning to artificial intelligence and machine learning providers, as well as to in-house solutions to exploit the enormous amount of data generated daily. The consequence? Better-informed, predictive decision-making that optimises asset performance.
“There has been a lot of investment in generating more granular, real-time data,” says Eric Bindler, research director of digital water at Bluefield Research, a US-based advisory firm focused on water infrastructure. “Data often comes on screen and then is never looked at again, falling into a black box. AI can be really helpful in looking back through these historical data sets, interpreting patterns and then finding insights that maybe one individual department or individual operator would not be able to come up with or process on their own.”
AI gets wet
In the water sector alone, the financial incentives and potential benefits of deploying AI are huge. In the US, the Environmental Protection Agency estimates that $473 billion is required to maintain and improve the country’s drinking water infrastructure, while the cost of modernising the wastewater sector has risen to over $271 billion.
Pre-pandemic, water infrastructure already required urgent attention, but volatile energy prices are increasing pressure on utilities and fuelling appetite for digital tools like AI. Operators of sewers and water treatment plants are turning to AI to ensure optimal energy-efficient performance, cutting costs without compromising safety standards.
“It is still pretty early days for AI in the water industry,” adds Bindler. “One of the main applications we see is helping customers to manage their consumption and detect leaks. If you unleash AI on the data, you can use it to identify patterns of normal consumption spikes and determine where anomalies might be leaks. AI can then alert customers or utilities directly, preventing wasted water and high cost to the consumer.”
Last year, the sale of US water infrastructure software firm Innovyze showcased the growing attraction of AI. Sweden-headquartered asset manager EQT offloaded the tech company for $1 billion, almost quadrupling the sum it initially paid out to acquire the business back in 2017.
Investment incentives for US water infrastructure were also strengthened in November when President Joe Biden signed the Bipartisan Infrastructure Law. The bill included a $50 billion investment pledge to modernise the country’s drinking water and wastewater systems.
Private investment is already starting to trickle in. In March, venture capital firm Emerald Technology Ventures tapped into its $100 million water innovation impact fund to invest in US infrastructure service provider SewerAI. The company uses CCTV imagery to develop AI and deep learning models that detect pipeline anomalies and faults in sewers, helping to cut costs and time-consuming manual inspections.
Fork in the road
Other infrastructure subsectors are also eyeing the benefits of AI. Across transportation, a dizzying number of potential use cases have already been mooted, from autonomous aircraft landing to traffic management.
Wintics, a French software firm partly owned by private equity firm Ardian, is a prime example of AI in action. The company deploys the technology to ease urban congestion and improve traffic mobility. “Wintics uses video streams from cameras that are installed on roads, airports and other forms of infrastructure,” explains Pauline Thomson, director at Ardian. “They have developed algorithms to detect abnormal behaviour on the road. For example, a car could go the wrong way on the highway, and they would be automatically detected.”
AI used in this manner helps streamline traffic flow, creating safer public spaces for pedestrians and promoting soft mobility. Traffic lights can be altered automatically to prevent bottlenecks and the controlled flow of vehicles is being trialled to promote greater bicycle use in urban areas. Previously, Wintics partnered with the city of Paris to measure how bike lanes could impact vehicle traffic, crucial for urban planning and decarbonising road networks.
Energy is also an important arena for AI, particularly renewables. “On our renewable energy assets, we are looking at predictive maintenance by examining historical trends to see if there are any material deviations,” says Thomson. “The limitation is, of course, that you can only predict the predictable, you can never predict something that has never happened in the past.”
Maintenance is a critical issue for reducing emissions across infrastructure subsectors. Extending the life expectancy of renewables assets, for example wind turbines, ensures more reliable energy supply, while a longer shelf life means less carbon-intensive material is required to replace defective equipment.
Canada-based software firm Clir Renewables is another company deploying AI in the renewables space. The firm combines advanced AI with the world’s largest renewable energy operational data set to boost operational efficiencies and improve performance.
“Aside from AI-powered insights specific to a single asset, such as predictive maintenance, power forecasting and component failures, machine learning and AI can be used on data sets curated from across the industry to provide insights into the relative performance and behaviour of assets,” says Mike Reid, CTO at Clir Renewables.
“For instance, we can use machine learning to analyse large amounts of supervisory control and data acquisition (SCADA) data to discover behavioural differences between original equipment manufacturers and component manufacturers during extreme wind speeds, which indicates a higher risk of failure for one asset over another.”
AI and other digital enablers will also be crucial for decarbonising carbon-intensive assets such as in the oil and gas sector. Increasing energy efficiency in refineries and upstream operations reduces emissions and methane leakage, a greenhouse gas more than 25 times more potent than carbon dioxide.
“You cannot just click your fingers and switch out conventional fossil fuels overnight, particularly considering how embedded they are in the global economy today,” explains Heptagon Capital’s Gunz. “Digital tools that can potentially accelerate that process or make it less painful are likely only going to be encouraged and supported.”
But AI is not a panacea for all operational ills. Much depends on the quality of the data and on clearly defining the task at hand.
“Availability of data is of utmost importance, as well as understanding what use case you want to solve for and the team to execute,” says Petter Weiderholm, EQT’s global head of IT strategy.
“This is typically where the lower overall digital maturity we see in traditional infrastructure assets makes it a little difficult to reap value from AI in the short term, as a lot of capabilities need to be built in the companies.”
Joe Perino, principal analyst at LNS Research, an advisory firm focused on industrial transformation, agrees that strategy needs to be front and centre. “AI is not the answer to everything. Most companies are not ready to scale AI up, [partly] because of a lack of skills [but also the] understanding of when and where to use it. Many execs get sold on this, spend $5 million on a platform licence and six months later they have nothing to show.”
The shift to greater digitalisation and third-party providers also increases the attack surface for potential malicious agents, as highlighted by last year’s high-profile ransomware takeover of the Colonial Pipeline in the US.
“Balancing cybersecurity is at the heart of many discussions when deploying technologies like AI,” adds Oleg-Serguei Schkoda, senior director at TCG Digital, a digital transformation consultancy. “The value of data is huge as [compromising cybersecurity] could lead to loss of value, production and competitive advantage.”
Schkoda explains that accessing assets remotely, made more common by the pandemic, increases the risk of a cyberattack by giving hackers more entry points. Cybersecurity protocols will likely need updating across the sector to prevent potential data breaches and maintain appropriate digital hygiene.
Terry Mason, director in cyber-risk and privacy at risk mitigation consultancy HKA, adds that the main risk from “next generation” technologies like AI, and efforts to digitalise, is “the speed at which these migrations are performed”, which “unfortunately is often combined with weak oversight and risk management practices”.
Another issue that many businesses struggle with is finding and retaining the data science talent needed to successfully implement AI. “Workers no longer stay 30 years [at a company], and transferring knowledge from the most experienced to the least is a real challenge,” says Perino.
“With the explosion of demand for AI experts across the broader tech industry, competition for talent is fierce,” adds Reid. “Not only is the ideal researcher an AI expert, but they must also have a deep understanding of [specific] industries and the data produced by assets to make sense of the vast amounts of data being generated.”
Tibor Schwartz, senior adviser (asset management) at QIC Global Infrastructure, agrees. “Infrastructure organisations have adopted digital technologies relatively recently, and it is not necessarily in their DNA to invest enough and bring in people with the right skill sets and capabilities in high demand, such as AI.”
Many businesses have yet to even appoint chief technology officers, showcasing the relatively nascent level of digital maturity across infrastructure. “There is a cultural battle before a technology one,” says Schkoda.
“Company culture and board mindset is key to success. This needs to come from the top, otherwise [digitalisation] will never be adopted in the long term.”
As with everything digital, you need the technological skill set to enact change, but also the long-term vision and knowledge of the sector to integrate AI holistically.
“It is critical to think backwards from the perspective of end-users and invite them to play an active and ongoing role in creating the AI roadmap,” says Weiderholm. “Not only will this ensure that what you are building will deliver value/solve their needs – it will also help you ‘demystify’ and build trust in the underlying AI.”
Tech to clear the decks
Managing waste is certainly not a new challenge, but the environmental consequences of not finding solutions to the vast amount of waste generated daily are increasingly profound.
Software firm Rubicon is combining AI and machine-learning principles to tackle the problem. “[We] utilise camera technology to identify contamination in waste bins, the condition of containers, road conditions and illegal dumping,” says Phil Rodoni, CTO at Rubicon. “Our AI technology uses these images, as well as location data and other signifiers, to determine recycling rates, as well as flag critical infrastructure issues to municipal governments.”
Beyond sorting waste in the most efficient manner, AI can be used to find the most environmentally friendly and safest routes for waste and recycling trucks. “The more we know about waste and how it is handled, the easier it will be to create smart infrastructure and a stronger platform for the circular economy,” adds Rodoni.