Insights
Technical deep dives, market analysis and project stories from across our capability areas.
Most of the good opportunities in energy assets are already known to the people running them — what's missing is a number a board can defend. Quantifying the value first, before building anything, is usually the faster path.
Most of a renewable project's grid connection sits idle for all but a few hours a year. A simple, transparent way to rank which stranded capacity is worth reclaiming — before committing to deeper dispatch and network modelling.
▦There's a shape that turns up in almost every capacity plan, and most people never notice what it actually is. The grid carries expensive peaking capacity so individual sites don't have to — which changes how you should size for your own peak.
▦Why the smartest storage operators borrow a thirty-year-old idea from hydroelectric dams — and why optimising against a single forecast quietly leaves money on the table.
▦Energy leaders on complexity, connection, and the value it unlocks. Like the blind men and the elephant, people across the energy ecosystem often see only their own piece of a complex landscape — leading to inefficiency, higher risk, and slower projects.
◆The UK C&I renewable market is accelerating. The buildings and land are there. The data exists. The question is who can find and qualify the best opportunities first.
◈Why the data already exists — and why nobody has connected it. The information needed to build a predictive solar-targeting model for commercial property is public, comprehensive and mostly free. The hard part isn't access. It's judgement.
◈Rolling Hurst tested as an early warning for Texas electricity spikes, in forty-eight different configurations — and (spoiler) it lost every one.
◈Wavelet-based dispatch for heterogeneous generation assets. For small fleets, conventional optimization handles coordination. For larger ones, it quietly stops working — and most operators don't notice until they're already paying for it.
▦Most renewable portfolios today are performing exactly as expected — at least on paper. So why is realised revenue still coming in below expectations?
◈Energy markets are built on a simple idea: if prices change, behaviour will follow. In theory it works. In practice, it often doesn't.
◈More and more, the work is orchestrated — automation pipelines, scheduled processes, optimization jobs, increasingly autonomous systems. So we ask a different question: are we building for users, or for systems acting on behalf of users?
</>Control logic in structured electricity markets. Battery arbitrage is often presented as a tariff problem. In practice, it is a control problem.
▦A manufacturing facility runs widgets 24/7. Electricity is cheap at night, expensive at peak. When should you run the machine, and at what rate, to minimise energy cost while hitting your production target?
▦The European Union's Net-Zero Industry Act is more than a regulatory framework — it's a clear signal that the economics of clean energy are entering a new era.
◈As the world transitions to cleaner, decentralised power, a major question emerges: how do we design renewable energy systems that are both cost-effective and reliable?
▦For years, companies have proudly claimed "100% renewable energy" — backed by certificates and contracts. But the reality is more nuanced.
◈Two leading firms join forces to bring AI, market expertise, and data-driven decision support to the fragmented US energy landscape.
◈Custom hardware isn't mysterious — the challenge lies in deciding when it's truly justified. Context, compliance and lifecycle shape smarter hardware decisions in modern energy systems.
▚For a century, balancing the grid meant building bigger power stations. A different model has now reached scale: thousands of small, distributed resources, coordinated to act as one.
▦Our team has been taking a close look at Ohio's private-wire legislation — Substitute House Bill 15 — and it stands out as one of the most imaginative and forward-leaning frameworks we've seen anywhere in the world.
◈Headlines celebrate record renewable capacity. Beneath them, a quieter and more uncomfortable trend: the firm, dispatchable capacity that keeps the lights on is retiring faster than it's being effectively replaced.
▦Energy DNA, zero drama. We're a full-stack engineering partner focused entirely on energy — you keep control, you keep the IP, and we add engineers who already know the domain.
◆The technology behind virtual power plants is mature and broadly the same everywhere. So why do VPPs flourish in some markets and stall in others? The answer is rarely the hardware. It's the rules.
▦On 15 July 2025, Ireland approved a landmark Private Wires policy — allowing privately owned electricity lines in specific, tightly defined cases. It's not a game changer on its own, but it's a clear signal that the tide is turning.
◈By combining cutting-edge BLE technologies, we help businesses not only comply with regulations but proactively enhance operational efficiency and sustainability — here, tracking modular office partitions through a circular-economy lifecycle.
▚AI has triggered the largest surge in projected electricity demand in a generation. The crucial question for everyone planning the grid is whether that demand is real and durable — or a bubble in the interconnection queue.
◈Our name might hint at software, but Full Stack Energy goes deeper. We engineer across the entire stack — from silicon to cloud — and custom hardware design is where some of the most transformative work happens. Because not every breakthrough fits an off-the-shelf mould.
▚When your timelines are tight, your market is complex, and the IP is yours to keep, bringing in the right experts isn't a luxury. It's common sense.
◆A quiet conviction: great engineering isn't about building the most — it's about building what matters, and no more. We call it Just Enough Engineering.
◆ETH Zurich, EPFL and CSC are deploying a fully open-source large language model — trained on Switzerland's "Alps" supercomputer and set to debut in late summer 2025.
ƒData centres are almost always described as a problem for the grid: vast, fast-growing, always-on loads. But that framing misses their most valuable and least-discussed property — much of that load can move.
◈Match every hour of your electricity use with carbon-free generation in the same region. As a sentence, it's simple. As an operating requirement, it's one of the hardest problems in corporate energy.
◈For a decade, the EV charging story was about rollout — more chargers, more locations, faster. In 2025 the defining question changed: not how many we can build, but whether the ones we have actually work.
⚡There was a time when logging was simple — a few print statements and a prayer that production wouldn't fall over. In the era of IoT, cloud scale and endless microservices, logs rain down like a biblical flood. The challenge is no longer collecting them; it's finding the signal in the noise.
ƒAlmost everything about how we think about charging an electric car is borrowed from the petrol station. That mental model is intuitive, familiar — and mostly wrong.
⚡The data that could answer your question usually exists. The barrier is that getting to it requires knowing where it lives and how to query it — skills most people in an organisation don't have.
ƒFor many commercial and industrial sites, the largest part of the electricity bill isn't how much energy you use — it's the brief moments when you use it fastest.
▦All over every building, devices that look switched off are quietly sipping power — standby loads that never stop. Individually trivial, collectively a real cost. And exactly the kind of pattern AI is good at hunting down.
ƒSwapping diesel vans for electric ones looks, on the surface, like a procurement decision. In reality it reorganises how a fleet is fuelled, parked, routed and financed — all at once.
⚡Buying the vehicles and installing the chargers is the project. Keeping every vehicle charged and ready, every single day, within the power you have — that's the operation. And the operation is where electrification gets hard.
⚡Reliable power used to be something you simply assumed. Increasingly, it's something you actively buy, design for, and pay a premium to guarantee.
▦Holding a vessel still in moving water sounds passive. For the battery bank doing it, it's one of the most punishing duty cycles there is — and a perfect lesson in why cell balancing matters.
▦Every EV driver knows the dance: a different app for each network, a new account, another stored card, and the dread that the one charger you've driven to needs a membership you don't have. The mess has roots deeper than bad apps.
⚡Home charging is easy for those who have a driveway. Public charging is where electrification has to work for everyone else — and it's where the experience most often breaks down.
⚡One EV charging is invisible to the grid. Millions charging at once — all plugging in at 6pm when everyone gets home — is a different story entirely.
⚡A digital twin used to be a mirror — a live model that showed you what was happening. Add AI and it becomes something more useful: a model that tells you what's about to happen, and what to do about it.
ƒThe energy transition is often discussed in terms of breakthroughs — the one technology that will change everything. Most of the real progress looks nothing like that.
⚡The move to electric transport is no longer a question of if, but of how fast — and how well the energy system adapts to it. Because an EV isn't just a cleaner car. It's a new node on the grid.
⚡Artificial intelligence is no longer a future promise in energy — it's quietly embedded in forecasting, maintenance, control and analysis. The opportunity is real, provided it's pointed at the right problems.
ƒAn AI model can be impressively accurate and still be unusable where it matters most — because in systems that run real infrastructure, "trust me" is not an acceptable answer.
ƒ"Level 1, 2 and 3" sound like a simple ladder of speed. They are — but the real difference is where the AC becomes DC, and that one detail decides everything about cost, speed and siting.
⚡Charging infrastructure either talks in open standards or it doesn't. That single choice decides whether your estate is a flexible, future-proof system — or a museum of incompatible hardware.
⚡A parked EV is a battery doing nothing. Smart charging decides when it fills; V2G lets it give back. Together they turn a fleet of cars into one of the largest distributed storage resources on the grid.
⚡"EVSE" is the technical name for the thing most people just call an EV charger. The acronym is more accurate — because the equipment does far more than push electrons down a cable.
⚡The Open Charge Point Protocol is the quiet standard that decides whether your charging estate is an open, future-proof system — or a pile of hardware you can never switch away from.
⚡Signed into law on 16 August 2022, the IRA is America's first-ever climate legislation — $369 billion in extended incentives to transition the US away from fossil-fuel energy, making the renewable sector more competitive by cutting construction costs.
◈It's often said in management circles that "if something can't be measured then it can't be managed." It can sound like a meaningless boardroom buzzword — but it's remarkably true when it comes to energy efficiency.
↗The ongoing Russian invasion of Ukraine has sent shockwaves across energy markets — most specifically in Europe, which depends on Russian energy imports for a high proportion of its energy needs.
◈A building's electricity consumption is rarely a single, simple signal. It's a composite — many overlapping patterns summed into one trace — and the value is in pulling them apart.
↗Forwarding data to InfluxDB via MQTT and Telegraf. One of the most common things I do when collecting time-series data from sensors deployed in the field — sometimes literally — is having them populate our time-series database of choice, InfluxDB.
▚If you're gathering instantaneous time-series data such as electricity usage in Watts over specific time periods, then at some point you're going to want to convert this into usage data — i.e. kWh or suchlike. It's less obvious than it looks.
↗The Irish electricity system is operated at a nominal frequency of 50Hz, with a normal operating range of ±0.2Hz. That frequency reflects the balance between system demand and generation — and read at high resolution, it tells a remarkable story.
↗Exponential growth isn't just confined to compound interest and capacitors. Using early Covid-19 case data, here's how the number of confirmed cases changed outside China as it spread across the world — and what the maths says next.
ƒHow a brewer at Guinness became "the Faraday of statistics" — William Sealy Gosset's work has proven fundamental to statistical inference as it's practised today.
ƒBack in July 2019, the German economy minister Peter Altmaier let slip that Germany "has a claim to digital sovereignty" and needs a European cloud industry of its own to rival US corporations such as Amazon. Why not?
</>When modelling the real world, it's surprising how many things I take for granted. It's worthwhile considering how seemingly obvious "given" information is actually derived — the process often sheds light on much harder problems.
ƒThe Fourier transform decomposes a signal into its constituent sine waves, and engineering wouldn't be the same without it. But before any of that, we need the thing it relies on: numbers — right up to the ultimate type, the complex number.
ƒImagine you're stranded on a desert island and you really need a good approximation of π for some super machine you're building to make your escape. You can't remember enough decimal places — but you do remember your fractals.
ƒThe aim of this post is to show how easy it is to do some very basic range testing with two LoRa devices — one operating as a transmitter and the other as a receiver.
▚A few people asked which SQL query I used to calculate the first-digit frequency distribution in the previous post on Benford's Law. Here it is — short and sweet.
ƒBenford's Law — the first-digit law — says that in many real-life datasets, 1 leads about 30% of the time and 9 less than 5%. Does it hold for energy data? Let's run the test.
ƒOccasionally we publish something about how we actually build our hardware and software. Here's a small algorithm we use to shape events that might otherwise overwhelm a network or a resource.
▚Click any insight to read it.