In his 2001 book, Winning with Software: An Executive Strategy, Watts S. Humphrey claimed that, “Every business is a software business.” Humphrey was a software engineering pioneer who foresaw that technology would soon (and very quickly) transform the way we work, so he was keen to let executives know they should prepare their companies for the inevitable.
Satya Nadella, CEO of Microsoft, gave the sentiment new life and added attention when he remarked that “every business will become a software business” – which works out nicely for companies that sell software, like Microsoft. But it’s not just software, and the phrase has been adapted to suit the needs of myriad marketers trying to position their offering as being on the cusp on ubiquity: “Every company is a media company,” “Every business is a data business,” “Every business is a digital business,” and so on.
So, based on the explosion of interest and investment in artificial intelligence (AI) this past year, is it fair to say that now every business is an AI business? For utilities, it’s far more complicated than that.
Buying vs. building an AI model
Utilities power lives and livelihoods, something they have been doing since Thomas Edison installed the world's first central generating station in 1882. There have been tremendous advancements since then, and utilities have managed to keep pace with innovation. We’ve now reached a whole new tipping point though, an age of accelerated digital transformation that has forced even the most traditional institutions to rebrand their relevance. But the fact remains that utility companies are still energy companies – they aren’t data companies, they aren’t machine learning companies, and they aren’t AI companies. And they shouldn’t have to be. After all, a digital transformation is about transforming your operations, not your identity.
As utility leaders begin to execute their 2024 digital strategies, now is the time to be crystal clear about the roles, responsibilities, and risks associated with digitizing such critical legacy systems. Does your team possess the technical expertise (and time) to take on such a complex overhaul? How might it impact your customers? And what’s the plan for the future? The urgency is there and the opportunities are clear, however many utilities are still vexed by the age-old question of ROI. Do they buy an AI model, or build their own?
Here are 3 things for utilities to think about as they decide whether to buy or build:
1. Understand the tech, the terms, and the time
When OpenAI released ChatGPT in November 2022, it was the first time most people had been introduced to generative AI, a version of AI that can “self-generate” text, images, or code. Fascinating to some and horrifying to others, generative AI isn’t necessarily what we’re talking about here. Rather, industry experts feel there will likely be two sets of use cases for AI in the utility sector – “more immediate, low-risk possibilities like using generative AI to post on social media” and “higher-risk AI functions related to a utility’s core activities like grid planning.” The higher the risk, the more regulatory oversight, so don’t expect sweeping changes overnight. While other service industries might be able to see some quick, flashy wins with their AI strategies, utility leaders must view it as an ongoing investment. And according to McKinsey, “bold [utility] industry companies that adopt a digital platform could achieve a step-change performance ahead of peers and, more important[ly], use the once-in-a-generation opportunity to fundamentally restructure the entire industry.”
2. Utility data is not in the public domain
AI is trained on data. Lots of data. For utility companies, this means inputting all the variables that could impact their assets – past performance, environmental threats, maintenance costs, load forecasting, etc. The resulting AI models must then be trained on this data, deployed, retrained, and monitored. If you’re a utility with a strong data-science program, then building your own AI model might make a lot of sense. But here’s the catch – due to obvious security concerns, utility data is not in the public domain. Meaning, while ChatGPT is a large learning model (or LLM) that can scour the entire public internet for knowledge, utilities building their own AI model are limited to learning from their own data inputs. This is why many companies opt to bring in an external vendor to help provide much deeper insights (i.e. more industry data) that will help build a more sophisticated AI model trained on more potential scenarios/complications.
3. Successful transformations don’t end, they endure
When you buy and install software, it serves its specific function. But when you buy and install (or build) an AI model, it keeps learning and growing. The lifecycle of software ends at installation, whereas the lifecycle of an AI model begins at installation. Instead of getting distracted by the hoopla of what other companies are doing right now in AI, utilities should adopt an outcome-driven methodology that begins with an honest conversation about long-term value. Otherwise, it'd be like moving to a foreign country and putting all of your efforts into planning the travel, instead of ensuring you’re prepared to learn the language and thrive in your new environment. The goal must always be sustainable transformation.
You don't have to start at the beginning
As daunting as a digital transformation might seem, there is an incredible amount of support available whether you ultimately choose to buy or build a digital or AI system. And it’s not an “all-or-nothing” decision. Just getting started? Look for those companies/consultants that might reduce the initial learning curve. Had a bad result? Bring on some extra resources until you’re ready to take the reins. It’s going to take more than one business to transform the utilities sector, and not every business is an AI business.
Vikhyat Chaudhry is CTO, COO and co-founder of Buzz Solutions.
