Attractiveness

Back in 2017, Nvidia’s CEO told us that “Software is eating the world, but AI was going to eat Software.” He wasn’t wrong. Last year alone, AI start-ups raised $73.3 billion in investment. The opportunities for asset optimizing are projected to skyrocket given that by 2030, the expected  number of connected devices and objects globally is forecast to reach 500 billion, up from around 50 billion today, according to Cisco.

Like many industries, the energy sector has been scrambling to keep pace with the opportunities AI presents and its inherent complexity means the sector is a prime example of how AI can eat the existing software used by utilities and grid operators today.

As of today, most utilities have looked to use AI for ways to improve what they already do in a more streamlined, automated fashion, often coming up short in envisaging how AI can rewrite the rule book. Many of the sector’s low-hanging fruit including system maintenance and business process have already been targeted.

However, interest in the next wave of AI innovation is accelerating. In 2020 energy AI start-ups raised $16.6 billion in venture investment.  Market stakeholders are beginning to consider more ambitious applications, often cross pollinating from other sectors such as manufacturing, health-tech and finance. While many innovators claim the use of AI as a marketing ploy, here are three areas advanced AI can really make an impact:

1. Grid Planning
2. Grid Inspection
3. Consumer Engagement

Grid Planning

You only have to back one month to Biden’s $8 billion transmission infrastructure announcement to understand the scale of grid expansion underway. Transformations incorporating an ever-increasing web of renewables, energy storage and electric vehicles requires increasingly complex models to simulate, model and optimize the best possible expansion strategy. AI is perfectly suited to such an application. Today’s grid planning market is largely represented by a bundle of various off-the-shelf products designed for yesteryear’s design methodology, or consulting services for individual projects.

With the ability to ingest vast amounts of data from a multitude of inputs to design, simulate and optimize dynamic multi-billion-node systems, the potential for AI in this market is largely untapped. Notable innovators in the space include:

• Envelio, a German-based start-up founded in 2017 has developed a platform for distribution grid operators to digitize energy planning and operation processes. By developing a dynamic data input system from their clients, they use machine learning and neural network modelling to develop future grid scenarios for short, medium and long-term planning out to 2030. With a strong foothold in their home market, the company have international projects in new locations including the United States and Brazil where regulators are enforcing a DER capacity map..
• Pearl Street Technologies has taken the method behind computer chip circuit design and simulation to grid planning, developing a physical based grid model which can run simulation and optimization scenarios for long-term planning. Initially targeting the transmission networks in the US is a smart move by the firm, given its federally regulated nature and therefore the ability to access clean data at a national scale. With seed funding raised in January, the group plans to increase its portfolio of ISO customers with a longer-term vision of becoming a one-stop shop for both for planning real-time control for all distribution, transmission, and microgrids networks.

Grid planning will benefit from such innovations. However, many existing operators,  grid  planners and regulators are not yet at the stage where they require such advanced tools, and this may be a case where the demand is not yet at the rate of innovation supply.  Another example is how utilities remain relentlessly slow at full-scale technology implementation. With other exciting R&D underway such as the EU Horizon 2020 FlexPlan project and Breakthrough Energy Sciences; US-wide grid scenario modelling tool, it’s a theme worth watching.

Grid inspection

The use of visual data inspection in the energy market has been widely adopted by a] forward thinking utilities, given its potential to drastically cut costs by removing on-site manual inspection.  Companies on the wildfire frontlines like PG&E are a prime example. As I covered in a previous insight following 2018 wildfires, the company was fined $25.5 billion on top of $13.5 billion for personal loss, resulting in bankruptcy in January 2019 (they since re-emerged following a supported $58 billion revenue injection). PG&E is now doubling down of advanced asset inspection technology. Last year the company inspected 15,000 miles of electric lines, using drones and helicopter inspection.

What role can the AI evolution play here? Innovators are looking to increase the quality of inspection through higher data fidelity to feed into more accurate condition monitoring tools, diversifying the range of data types that can be used in AI models, and lowering the cost of inspection further through low-cost hardware. Notable innovators include:

• Cogniac, a Silicon Valley born start-up with an AI-SaaS product to automate visual inspection tasks utilizing neutral networks to continuously improve image-agnostic condition monitoring.  They currently insect anywhere between 100 to 50 million client assets per month, covering a broad array of infrastructure. With $20 million raised to date the company is growing radially internationally.
• Cyberhawk a UK-based provider of aerial inspection and surveying services using drones. SP Energy Networks (SPEN) has been a key customer since the beginning who has now fully switched to drone technology for transmission network inspection. Where the company differentiates <itself? >is with the iHawk platform, a source-of-truth data management platform for clients to ingest any form of data for global dynamic asset condition monitoring.

Data quality is the core challenge plaguing many of the efforts today. Most utilities lack imaging technology with sufficient resolution, which can require significant capital investments. Therefore, smart approaches to sensor/impacting financing will be worth considering in partnership with AI providers in the market.

Consumer engagement

It’s well known that most consumers would prefer to not think about where their energy comes from, or how it’s used or where it goes. The energy sector is an industry where end-customer centricity is not always apparent in the core products sold. The opportunity therefore is for AI to overcome these industry quirks, while providing value to a utility by enhancing customer lifetime value and helping to unlock potential future consumer investment into grid edge opportunities.

As we covered here, consumer appliance data disaggregation (I.e., appliance end-use data) has been a tried and tested approach by many utilities and energy retailers and is seen as a popular applications for AI tools to provide insights on user behavior. Today, the trends for increasing consumer engagement using AI involve increased data granularity using real-time data, on-site edge computing data processing and service upselling, using AI to open the door to additional services including EV charging and boarder home energy management. Innovators in the market include:

• Uplight, a market leader in US consumer engagement, serving over 85 utilities across the US and valued at $1.5 billion. The company istransiting from a consumer engagement utility white-label offering to a holistic energy management service which can enable demand-side flexibility management (covered here), e-mobility and others.
• Net2Grid– a Netherlands-based start-up with an energy insights SaaS platform for disaggregation meter data. The company is differentiated in the market using real-time data which is being used for inter/intraday market DER forecasting and real-time data processing on sub-meters. The company closed a Series B round this March for international expansion, just opening a US office.

With a slow transition towards time-of-use tariffs, real-time data is the next big trend in the space, enabling an array of service opportunities. The key challenge here remains security, as many utilities choose not to put meter data on the cloud, limiting the value you can add to the local environment.

Keep an eye on

• With the plethora of AI solution providers offering an ever increasing array of applications, keep an eye out for companies such as the Enovation Hub which is focused on connecting utility needs to AI solution providers.
• Briefly mentioned, the role of edge computing for meter-data processing to overcome security and data latency issues Read more about edge computing.Quantum computing as the next wave of technology to impact some of the applications mentioned above Read more about quantum computing. (see another article here).