What AI Actually Means for Energy Analysts Right Now

ELI5/TLDR

The CEO of Modo Energy — a company that sells forecasts and benchmarks for battery storage projects — closes his conference with a pitch about AI. His core claim: analysts spend 80% of their time pulling numbers out of spreadsheets and reformatting reports, and only 20% on the judgment work that actually matters. Modo has built an AI assistant called Co that aims to flip that ratio. It’s part victory lap (the company’s numbers are good), part product launch, part conference applause line.

The Full Story

The industry got bankable, and Modo rode the wave

The first half is a state-of-the-union. Battery energy storage — building giant batteries that store electricity and sell it back to the grid when prices spike — has gone from “fit the whole community in the back room of a pub” to a room of 800 people with another 800 turned away. The reason, Scrimshire argues, is that batteries became a bankable asset class: something serious money managers will lend against and own.

The supporting numbers: 87 gigawatts of storage added globally last year, an expected 130+ in 2026. £700 billion flowed into renewables and storage last year, projected to hit £1.3 trillion by decade’s end. Modo’s own report card is the subtext throughout — 200 subscribers, 30,000 terminal users, live in 15 regions heading to 25 by 2027, £3 billion of assets financed using its forecasts this year with £10 billion targeted next.

Really the only thing that matters is that we’re delivering shareholder value. And that’s the thing that keeps the flywheel spinning so we can make more bets and allocate more capital.

Two things to decode. A benchmark is a trusted reference price — like a stock index, but for what a battery earns — and Modo says it’s now the only FCA-regulated one in the world (FCA being the UK’s financial regulator). That regulatory stamp matters because once a price is officially trustworthy, people build financial contracts on top of it: swaps, insurance, tolling agreements. A forecast, separately, is Modo’s projection of what an asset will earn over its life, which lenders use to size a loan.

”The consultancy era is over”

The pivot. For decades this industry leaned on consultancies for valuations and market intelligence. Scrimshire’s complaint: black-box models you can’t inspect, slow cycle times, everything trapped in Excel sheets and PDFs, expensive. Modo’s first act was to automate and standardize some of that, transparently. But he frames that as incremental. The real change is AI.

Certainly since the models came out from Anthropic in December, the whole world has changed.

He credits three shifts: better foundational models, better context management, and — the one he leans on — agents, meaning AI that doesn’t just answer a question but goes off and does multi-step work on its own.

Co, and the 80/20 flip

The product is Co, an “AI energy analyst” Modo has built over 18 months, released to subscribers in beta after Christmas. The framing is the 80/20 split: the busywork was never the job, the judgment was. Co today answers questions grounded in Modo’s data. The Co they’re building is more ambitious — give it a task, it writes a plan, runs the forecast, does the analysis, drafts your investment-committee paper, and checks its own work.

The adoption stats are the most concrete claim in the talk: three months after launch, of 30,000 terminal users, 25% are weekly-active on Co and 20% daily-active. They’ve also shipped MCP (a standard plumbing layer that lets the AI pull data straight into Excel, Word, and PowerPoint), with the UK and Germany leading usage.

On the jobs question he takes the optimistic line — hiring more analysts than ever, and reframing internal roles: analysts are becoming “power market engineers” whose main job is training Co. The team reportedly spends more than half its time training the AI and is the product’s biggest internal user — eating the dog food, as he puts it.

The proof and the giveaway

A polished product video plays, then real customer quotes (“answers that took days from a consultant, Co gives in seconds”). The close is a one-month removal of all usage limits on Co for paying customers — “drive it like you stole it” — pitched as a way to push token usage and gather feedback. Then everyone goes to the bar.

Key Takeaways

• Global battery storage: 87 GW added in 2025, 130+ GW expected in 2026. Capital into renewables and storage: £700B last year, £1.3T projected by 2030.
• Modo claims to be the first and only FCA-regulated provider of battery benchmarks — which is what lets swaps, tolling, floor agreements, and insurance underwriting get built on top.
• The pitch’s spine: analysts spend 80% of time on busywork (spreadsheet extraction, reformatting board packs), 20% on judgment. AI is sold as the lever to invert that.
• Co adoption (their headline metric): 25% of 30,000 terminal users weekly-active, 20% daily-active, three months post-launch.
• “The consultancy era is over” — the explicit thesis is AI displacing the legacy energy-consultancy model of black-box Excel-and-PDF deliverables.
• The aspirational agent does end-to-end work — plan, forecast, analyze, draft the IC paper, self-check — but he’s clear it isn’t there yet.
• Modo says its AI is held to the same governance and audit standards (Bloomberg, S&P, MSCI, Reuters) as its regulated benchmarks — trust as the moat.

Claude’s Take

This is a conference closing keynote, which means it’s a sales pitch wearing the clothes of an industry talk. Read it that way and it’s fine — even useful. The genuinely interesting signal is the adoption data: 20% daily-active on a three-month-old AI tool inside a niche professional terminal is a real number, not vapor, and it’s a clean data point on how fast agentic tools are landing in unglamorous B2B workflows. The energy-storage macro figures are also worth filing.

The soft spots are the usual keynote moves. “The consultancy era is over” is the kind of line that gets applause at your own conference and is much harder to defend in the room next door. The 80/20 framing is a real insight but also the single most-recycled slide in every AI-productivity pitch of the last two years. And the “we’re hiring more people, not fewer” reassurance sits a little awkwardly next to “we imagine a world where a lot of the work in this room is done by agents” — both can’t be load-bearing forever. The aspirational Co that drafts and self-checks IC papers is described in future tense for a reason.

Nothing here is dishonest; it’s just that the persuasive parts and the informative parts are tangled together, and the talk ends — literally — at the bar. A 5: solid as a snapshot of where AI sits in one real industry, low on anything you couldn’t predict from the genre.