Knowledge Relay

08/27/2026

AI doesn't improve just one phase of outage management.

Done right, it reshapes the entire lifecycle — from the first risk signal to the last lessons-learned entry.

Here's what an AI-native outage looks like, stage by stage:

1️⃣ Pre-Outage — Predict & Prepare
ML models flag vulnerable equipment weeks ahead. Scope that would have been a mid-outage discovery becomes a planned work package instead.

2️⃣ Planning & Scheduling — Automate & Stress-Test
Generative AI drafts initial scope and schedules. Monte Carlo simulation builds probabilistic risk overlays — identifying fragile sequences that fixed-point planning misses entirely.

3️⃣ Ex*****on — See & Act in Real Time
AI-powered dashboards give outage managers live visibility into progress, constraints, and emerging risk. Decisions happen in hours, not the next morning's stand-up.

4️⃣ Post-Outage Learning — Improve the Next Cycle
Every task duration, emergent work order, and deviation feeds back into the models. Each outage makes the next one smarter.

The shift from manual to AI-native starts with one decision: unifying your outage data into a foundation AI can actually learn from.

👉 Read the full executive playbook: knowledgerelay.com/blog/executive-playbook-using-ai-to-shorten-outages-at-nuclear-power-plants/

08/25/2026

Manual vs AI-driven nuclear outages — the difference is bigger than most people realize. 👇

Across four critical KPIs, here's what separates the plants that run predictable outages from the ones that don't:

🕐 Outage Duration — Manual: longer and less predictable. AI: shorter and on plan.
🔧 Emergent Work — Manual: high surprises mid-outage. AI: minimized surprises, caught pre-outage.
📋 Planning Effort — Manual: siloed and slow. AI: automated and collaborative.
📊 Decision Latency — Manual: reactive. AI: real-time and proactive.

It starts with a unified data foundation. That's what gives AI something to actually work with.

👉 Learn more about the KR Data Lakehouse®: knowledgerelay.com/products/kr-data-lakehouse/

08/21/2026

The nuclear industry just got a major signal on where the workforce is headed and it validates exactly what we've been saying.

Atomic Canyon, backed by investment from NVIDIA and former Vanguard CEO Tim Buckley, has launched NIVA (Nuclear Industry Virtual Assistant), a generative AI tool built with Institute of Nuclear Power Operations (INPO), EPRI, and the Nuclear Energy Institute to help plant workers navigate decades of operating records and technical knowledge faster. Constellation Energy, the largest nuclear fleet operator in the country, is already using it.

CEO Trey Lauderdale put it bluntly: "I do not see any way the nuclear renaissance...could possibly achieve our goals if we do not leverage generative AI applications to enable the expansion of the capacity of our workforce."

That line should stop every plant leader, outage manager, and project controls professional in their tracks.

Why this matters right now:
▪️ Nuclear is enjoying its best growth outlook in decades; uprates, restarts, and new builds are all moving forward simultaneously.
▪️ That growth is colliding with a real technical talent shortage. There simply aren't enough engineers and experienced staff to do things the old way.
▪️ AI isn't positioned to replace the workforce, it's positioned to multiply its capacity. NIVA is built to help people find answers in massive document sets faster, not eliminate the people who use them.
▪️ This is now a fleet-level, industry-endorsed tool, not a pilot. That's a strong market signal that AI-assisted knowledge and operations tools have moved from "interesting" to "expected."

Atomic Canyon is solving for knowledge access and operating experience. Knowledge Relay solves for something adjacent and equally urgent, outage and project controls ex*****on.

The same workforce gap driving demand for NIVA is driving demand for smarter, AI-enabled scheduling, risk visibility, and deliverable tracking across outages and capital projects.

The plants that will win in this renaissance aren't just the ones with access to operating history; they're the ones that can execute flawlessly under compressed schedules with fewer experienced hands on deck. That's the gap Knowledge Relay is built to close: giving project controls teams the same kind of AI-augmented leverage on ex*****on that NIVA is giving them on knowledge retrieval.

The message from Nvidia's investment, EPRI's involvement, and Constellation's adoption is the same message we've been delivering to the fleet: AI-enabled tools are no longer optional infrastructure for nuclear, they're the only way the workforce math actually works.

The launch could boost industry AI use to maintain and expand nuclear energy.

08/21/2026

Every extra outage day costs nuclear operators more than $1 million. 💡

That's not an exaggeration — it's the cost of lost generation, contractor overtime, and replacement power when outages run longer than planned.

Here's what the data shows:
▶ Refueling outages can eat up to 25% of a plant's annual O&M budget.
▶ AI tools deployed at Constellation plants are saving millions per reactor each year — with $80M/yr projected if scaled across the full U.S. BWR fleet.
▶ Avoiding just 1–2 extra outage days can pay for an entire AI investment program.

We've put together a 4-panel infographic series breaking down how AI and data are changing the outage equation for nuclear operators. First panel is above — more coming each week.

👉 Explore KR AI and Machine Learning: knowledgerelay.com/products/kr-ai/

08/20/2026

The grid doesn't have a clean energy problem. It has a reliability problem.
Reuters laid out the case this week in "The little reactors that could? How SMRs became nuclear's best bet" but the headline story isn't really about reactor size. It's about demand outrunning everything intermittent power can deliver.

Here's the number that should be in every utility boardroom right now: AI data centers, industrial reshoring, and electrification are adding load to the grid faster than wind and solar, paired with storage, can reliably absorb it. U.S. Department of Energy (DOE) has said this explicitly: rising demand is the reason to move faster on nuclear, period.

Nuclear is the only carbon-free source that runs at 90%+ capacity factor, day and night, regardless of weather. That's not a talking point. That's the physics of baseload power. And it's exactly what a 24/7 data center or a reshored manufacturing plant actually needs to sign a power purchase agreement.

That's why SMRs are getting real permits and real construction now, TerraPower's Natrium project in Wyoming being the clearest example, and why the ADVANCE Act and the U.S. Nuclear Regulatory Commissions new Part 53 framework exist at all. Policy is finally responding to a demand curve, not the other way around.

But here's what doesn't get enough airtime: building the reactors is only half the equation. Every gigawatt of new nuclear — SMR or otherwise — has to be delivered on schedule and on budget to actually meet the demand curve it's chasing. Nuclear's biggest historical risk was never physics. It was cost and schedule discipline across long, complex builds. That risk doesn't disappear because the reactor got smaller.

The demand for reliable power isn't a future scenario anymore. It's already here. The question is whether the industry can deliver capacity on the timeline the grid actually needs.

08/18/2026

2026 is the year nuclear moved from a long-term energy-policy conversation to a core business and infrastructure priority.

The real story is nuclear is becoming an ex*****on strategy. Utilities, technology companies, policymakers, manufacturers, and investors are converging around the same challenge how to deliver firm, scalable power fast enough to support the next era of economic growth.

The reason is increasingly straightforward: electricity demand is rising faster and with more urgency than many grids were built to accommodate. AI, data centers, advanced manufacturing, electrification, and energy-security needs have pushed dependable, around-the-clock generation back to the center of the energy agenda.

Capital is responding. Nuclear M&A activity reached $7 billion in the first half of 2026, double the level of the prior-year period. The focus is no longer limited to reactor designs: investment is spreading across the fuel cycle, specialized manufacturing, engineering, logistics, and the broader supply chain required to execute at scale.

Importantly, this is not just a developer story. Large utilities are increasingly central to the next phase of nuclear deployment.

Tennessee Valley Authority is advancing the Clinch River SMR project, while its power-purchase arrangement is intended to connect Kairos Power’s HERMES II output with a Google data center. Rocky Mountain Power is also positioned to receive up to 500 MW from TerraPower’s Natrium project in Wyoming.

Established fleet operators, including Constellation and Dominion Energy remain essential to the equation as they preserve existing zero-carbon generation while the market develops next-generation capacity.

Meanwhile, companies such as BWXT, NANO Nuclear Energy Inc. (NASDAQ: NNE), Oklo Inc, and Westinghouse Electric Company are helping build the industrial base necessary to turn projects into deployable infrastructure.

The first half of 2026 has shown that nuclear is no longer a future-facing conversation, it is becoming a present-day ex*****on priority. What developments do you think will define the next chapter: utility commitments, new project starts, supply-chain scale-up, or a breakthrough in how large power customers procure clean, firm energy?

https://www.nei.org/news/taking-the-investment-pulse-h1-2026

08/15/2026

Exelon’s latest data-center load update is a timely reminder: the energy transition is not just a capacity-planning challenge, it is a project-controls challenge.

As utilities distinguish committed load from speculative requests, disciplined controls become essential: clear scope definition, credible schedules, cost accountability, risk management, and decision-quality data. Exelon’s use of transmission security agreements is one example of converting interest into measurable commitment while helping protect existing customers.

As demand forecasts drive investments in transmission, storage, and generation, success will depend on more than ambitious GW targets. It will depend on the ability to govern complex portfolios from forecast through ex*****on.

The drop to 11 GW reflects Exelon’s ability to w**d out speculative projects through “transmission security agreements,” said CFO Jeanne Jones.

08/13/2026

US Electricity Demand Growth 2027 Marks a Regime Change

US electricity demand growth in 2027 caps a milestone stretch: the country is entering its strongest four-year electricity demand growth stretch since 2000, with growth of 1.9% in 2026 and 2.5% in 2027, according to the U.S. Energy Information Administration.

A separate EIA press release puts 2027 growth at 3%. ERCOT is growing near 10% a year and PJM Interconnection near 3% a year through 2027. For anyone who managed outages against flat or declining load, this is the real regime change, not the AI headlines.

Flat load let utilities defer generation decisions and treat nuclear capacity as a known quantity. Sustained multi-percent growth forces a different posture: every megawatt of uprate, life extension, or restart becomes part of resource adequacy.

👉 What does 3% growth do to your next planning cycle?

https://www.eia.gov/todayinenergy/detail.php?id=67344
https://www.eia.gov/pressroom/releases/press582.php

08/12/2026

Governor Gavin Newsom says he will not extend operation of nuclear power plant (Diablo Canyon Power Plant) beyond 2030

The future of California’s only nuclear power plant is up in the air. Diablo Canyon is currently set to stop operating in 2030.

Governor Gavin Newsom recently addressed the plant’s future at a press conference in Oakland.

“Well, we just finished the process that unfolded over the course of the last 4 or 5 years, just to extend its life," Newsom said. "That will be the baton that I will pass on to the next administration. So that will be the determination of the next administration.”

The Federal Nuclear Regulatory Commission has already approved operations through 2045.

https://www.ksby.com/avila-beach/governor-gavin-newsom-says-he-will-not-extend-operation-of-nuclear-power-plant-diablo-canyon-beyond-2030

They can't close Diablo Canyon unless they want regular blackouts in Central and Northern California. Cities that will experience regular blackouts will include: San Francisco, San Jose, Oakland, Sunnyvale, San Luis Obispo, Santa Barbara, Paso Robles, Fresno, Bakersfield, and Stockton. This is an obvious blunder, as the rest of America is reopening plants such as Palisades, Michigan (Holtec International), Three Mile Island, Pennsylvania (Constellation), and Duane Arnold, Iowa (NextEra Energy, Inc.).

Solution: Tennessee is building a plant to recycle nuclear waste. France is already doing it. New Small Modular Reactors can use the waste as fuel. This problem already has a solution. So why are we closing California's best source of reliable base-load energy?

The professionals working there for Pacific Gas and Electric Company are doing an amazing job. These kinds of failures of leadership cause excellent resources to leave California, fearing their careers will be ended if they don't.

We need to contact the governor's office to rethink this huge miss.

08/11/2026

The U.S. power market is entering a new phase: reliability assets are being repriced and repositioned.

Energy’s agreement to sell the 606 MW Brazos Valley Energy Center in ERCOT to LS Power for $860 million is more than an individual asset transaction. It is another signal that proven, dispatchable generation has become strategically important as electricity demand accelerates.

The distinction matters: this is not a physical conversion of a gas plant. It is an ownership and operating-model transition—placing an existing, efficient combined-cycle asset with an owner whose strategy is focused on optimizing flexible generation capacity in one of North America’s fastest-growing power markets.

For Constellation, the divestiture supports the completion of its Calpine transaction and further reshapes its broader portfolio. For LS Power, it adds reliable, around-the-clock capacity at a time when new generation and transmission projects can take years to permit, finance, and construct.

Across the industry, we are seeing a clear trend:
🔸 Existing gas-generation assets are being acquired, retained, upgraded, and operationally optimized rather than simply retired.
🔸 Data centers, AI infrastructure, electrification, reshoring, and population growth are materially changing load forecasts.
🔸 Grid planners and power providers need resources that can respond when variable generation is unavailable, transmission is constrained, or demand spikes.

The energy transition is increasingly an “and” conversation—not an “either/or” conversation: nuclear, renewables, storage, demand response, transmission, and dispatchable thermal generation each have a role.

The strategic question for utilities, generators, regulators, and large-load customers is no longer whether demand growth is coming. It is whether the industry can bring dependable capacity online—or preserve it—in time.
In that environment, the value of existing, well-sited, dispatchable plants is likely to remain central to both grid reliability and economic development.

https://www.reuters.com/business/energy/constellation-sell-gas-plant-ls-power-raises-operating-earnings-forecast-2026-08-06/