Aalo Atomics sees 92% cut in permit time with Azure, saving $80M yearly

When he was a child in Bangladesh, evenings could go dark without warning, recalls Yasir Arafat, Co-founder, President, and CTO at Aalo Atomics. Families lit candles while they waited for the power to return. This scene mirrors a global scenario unfolding today as rising energy demand exceeds what current systems can reliably support. Hyperscale computing, industrial electrification, and aging grid infrastructure all contribute to a widening gap between what’s required and what today’s energy supplies can deliver. 

Nuclear power offers reliability and scale, but deploying new reactors remains slow. Licensing alone can take years, requiring thousands of pages of documentation, in addition to revision and review cycles that span months.

Aalo intends to manufacture hundreds of nuclear reactors, each the size of a one-car garage, per year. But nuclear timelines are constrained by permitting processes that can’t keep pace with demand. Aalo, however, has a strategy for speed.

Scaling nuclear energy through repeatable permitting

“We are moving from treating nuclear power plants as a project to treating them as a product,” Arafat explains. “Instead of building one big power plant that may take more than 10 years, we can build a system to manufacture multiple smaller power plants, using a framework and infrastructure that lets us go fast.” 

To create a fleet of nuclear reactors, Aalo needs a licensing and permitting foundation that scales alongside manufacturing, supply chain, construction, and plant operations. Microsoft Generative AI for Permitting, an AI-powered solution on Azure, acts as a force multiplier for engineering and licensing teams. 

“Two things matter most: enterprise-scale complexity and business-critical reliability. We’re deploying something complex at a scale only a company like Microsoft really understands. There’s no room for anything less than proven reliability,” explains Arafat.

The system drafts foundational documents, identifies gaps, connects requirements back to regulations, and updates the documents as engineering and design changes happen. A customization pipeline allows organizations to tailor the solution to a specific need another energy producer might have, such as wind or solar farm permitting. 

The solution drafts the foundation of a 2,000- to 3,000-page permit application in seconds, highlighting where more information is needed, high-risk scenarios, or areas of opportunity. “I put in a limited set of data points, click a button, and Generative AI for Permitting generates a report that would’ve taken literally years before. It has high integrity. If it doesn’t have the answer, it doesn’t make something up. That’s incredibly important when documentation must stand up to strict regulatory review,” says Christine Riegler, Head of Project Development at Aalo Atomics.   

Extending Generative AI for Permitting across operations

With a repeatable, fast permitting foundation, the company can begin connecting adjacent workflows across procurement, manufacturing, quality assurance, and operations. When permitting no longer sets the floor for cycle time, every step can move more quickly.

“Generative AI for Permitting gave us a high-performance foundation. We gained access to the underlying architecture and repository, which allows us to extend it across the other AI systems we’re building,” says Robert Kessler, Head of AI at Aalo Atomics. Aalo is learning best practices for its own AI development from Generative AI for Permitting and its ability to reason across large volumes of regulatory and technical information. Some custom use cases that Aalo developed on top of the system include a site selection AI agent and environmental reporting. An agentic workflow for analyzing purchase orders reduces human workload by 85%.

Aalo is building its manufacturing program around an AI observability stack on Azure. This gives the team real-time visibility into how designs, production data, and licensing updates propagate across the deployment lifecycle. 

Results that shape an operating model

What began with a permitting bottleneck is now shaping an operating model. With Generative AI for Permitting, Aalo anticipates a 92% cut in the permit timeline, from four years to four months. It also expects to save $80 million annually in licensing engineer costs for permit drafting once Aalo is manufacturing hundreds of reactors a year​. Using the system, Aalo creates environmental reports 50 times faster, in hours instead of weeks.

For Arafat, the work traces back to those candlelit nights in Bangladesh. Reliable power changes what communities can plan for, build, and become. By turning permitting into a living, scalable system, Aalo is working to bring abundant, clean electricity online fast enough to meet the moment.

Discover more about Aalo Atomics on LinkedIn, X/Twitter, and YouTube.