Playback speed

Share post at current time

Share from 0:00

0:00

What MaintainX’s $3.6 billion acquisition tells us about industrial AI and the value of solving boring problems.

The Physical World Has a Software Problem Before It Has an AI Problem

Devansh and Nick

Nov 09, 2026

In Silicon Valley, we’re talking about autonomous agents, world models, robotics, and machines that might eventually reason about the physical world. Meanwhile, many factories are still trying to get their maintenance teams off spreadsheets.

And there’s a surprisingly large business in fixing that.

I recently spoke with Nick Haase, co-founder of MaintainX, a company that builds software for the people maintaining factories, machines, and other physical infrastructure. MaintainX now serves roughly 15,000 organizations, according to Nick, and was acquired by Autodesk for approximately $3.6 billion in August 2026.

What interested me wasn’t simply the acquisition. It was how much of MaintainX’s success came from recognizing problems that look almost embarrassingly simple to someone working in modern software.

Factories are full of expensive machines, highly skilled workers, and sophisticated engineering. Yet much of the information needed to operate those machines is scattered across incompatible software, paper records, and the memories of people who have been doing the job for decades. As AI becomes increasingly capable, that gap is becoming more important. The opportunity isn’t necessarily to introduce the most advanced technology into a factory. It’s to understand why the factory isn’t already working as efficiently as it should, then figure out which parts of that problem technology can actually solve.

Community Spotlight: Robotics' New Era After Astra

Recently, I came across an article called Robotics' New Era After Astra by FD and it had one of the most surprising learnings I’ve read recently— “On September 4, RoboCurve published a simple experiment: OpenAI’s GPT-6 Astra controlled two robotic arms and was asked to pick up a block and place it into a bowl. It succeeded in 19 out of 20 trials. A month later, NVIDIA’s Astranomy study tested Astra across 120 simulated manipulation tasks. Its best configuration achieved a 74.8% success rate, significantly outperforming several specialized robotics models on the same benchmark.

The surprising part? Astra was not designed as a robotics foundation model. A general-purpose model trained on reasoning, vision, coding, and tool use can now perform meaningful physical tasks without task-specific robot-policy fine-tuning.”

I get a lot of questions about robotics and physical AI, and while I find the space interesting, I’m not an expert. I found this article to be an exceptional primer to this topic, with a great seeding for my further exploration. If you’re looking to learn more about all of this, this will be a great place to start.

Hey there. I keep this newsletter ad-free, w/o sponsors, and mostly w/o paywalls to ensure that everyone has access to high-quality deep-dives on the most important ideas in the AI Industry without any bias. B/w my research, independent verification, and the chocolate milk needed to produce these works, these deep-dives take a lot of time, resources, and infrastructure. For just 10 USD/month or 100 USD/year, you can help me continue my vision of democratizing access to the most important ideas in AI by becoming a paid subscriber over here —

Help me buy chocolate milk

(If that is too much for you, we follow a pay-what-you-can plan over here, and it goes down all the way to 1 USD/month or 10 USD/year. No matter what plan you pick, you will get all of our articles without paywalls and have access to all of our articles. Check it out here).

Companion Guide to the Livestream

This guide expands the core ideas and structures them for deeper reflection. Watch the full stream for tone, nuance, and side-commentary.

1. Manufacturing has a lot of basic problems left to solve

Nick estimates that maintenance technicians spend only 25–35% of their time actually working on equipment. The rest goes toward finding information, paperwork, coordination, and other tasks.
A big reason is accumulated software and processes. Companies introduce new systems without retiring old ones, leaving workers to manage information across multiple platforms.
Manufacturing is also relatively risk-averse, often for good reason. A failed software implementation can disrupt production and cost substantial money.
Many facilities still rely heavily on spreadsheets, whiteboards, and paper. Even reliable connectivity and mobile devices aren’t universal.

2. Industrial AI needs context, not just more data

Factories already generate large amounts of machine and sensor data. What they often lack is the context needed to interpret it.

Maintenance history, technician observations, equipment failures, and operational decisions frequently aren’t connected to that data.

This is also a workforce problem. Experienced technicians are retiring, taking years of undocumented knowledge with them.

Nick’s view is that the manufacturers that learn to capture and use this knowledge will have a substantial advantage. More sophisticated AI won’t accomplish much if the underlying information is unreliable.

3. MaintainX’s growth offers an interesting startup playbook

MaintainX didn’t begin by selling exclusively to large manufacturers. Some of its earliest customers were country clubs, car washes, and fast-food franchises.

The company offered a free product and made it easy for smaller businesses to adopt its software.

That helped in several ways:

Smaller businesses had fewer procurement barriers and could adopt products faster.
Their feedback helped MaintainX improve its product.
The company discovered that maintenance problems were surprisingly similar across very different industries.
It could establish a customer base and credibility before pursuing major enterprise contracts.

Nick mentioned that some enterprise RFPs still ask whether a company has existed for at least ten years. Building trust takes time, and smaller customers can help bridge that gap.

4. The opportunity isn’t limited to reducing costs

We discussed the distinction between improving efficiency and expanding markets.

For a manufacturer operating at capacity, increasing production might normally require building another factory.

But if better maintenance and operations can increase throughput using existing equipment, the company can serve additional demand without that capital expenditure.

Nick also mentioned factories holding tens of millions of dollars in spare-parts inventory without adequate visibility into what they own.

These are opportunities to improve working capital, increase production, and make better use of existing assets.

5. More sophisticated AI isn’t always the right solution

I shared an experience working with a manufacturer on feasibility assessments for incoming customer orders.

Our initial instinct was to use a complex multi-agent system. But that created too much complexity for the users.

A simpler approach worked better: present the assumptions, explain the reasoning, and let the user question and refine the assessment.

Nick made a similar point about introducing AI to factory workers. Sometimes it’s better not to present a feature as AI at all. Just make it useful and understandable.

6. Autodesk and the industrial data opportunity

One interesting aspect of Autodesk’s acquisition of MaintainX is the potential to connect equipment design with actual operational performance.

Information about how machines fail, how they’re repaired, and how they perform in production could eventually inform future engineering decisions.

This creates a feedback loop between designing equipment and understanding how it behaves in the real world.

It’s also an example of why companies that capture useful operational data can become strategically valuable.

7. Advice for anyone building in industrial AI

Nick’s advice was remarkably practical: visit factories.

Call manufacturers, speak with production managers, attend regional trade shows, and ask to observe operations.

Don’t start by asking people what software they want. Ask what their day looks like, where they lose time, and which parts of their work are frustrating.

For someone early in their career, he even suggested spending a year working inside a manufacturing company and building internal tools.

His broader advice was to focus on solving one specific problem exceptionally well rather than trying to automate everything.

8. Where industrial AI goes next

Nick sees opportunities in material science, process simulation, equipment optimization, and predictive maintenance.

But much of the immediate opportunity is less complicated: improving basic information systems, reducing administrative work, and helping manufacturers make better decisions with the equipment and data they already have.

A recurring theme throughout our conversation was that technical capability and practical judgment are different things. AI is making it much easier to build software, but knowing what should be built still requires understanding the people and problems involved.

Subscribe to support AI Made Simple and help us deliver more quality information to you-

Flexible pricing available—pay what matches your budget here.

Thank you for being here, and I hope you have a wonderful day.

Dev <3

If you liked this article and wish to share it, please refer to the following guidelines.

That is it for this piece. I appreciate your time. As always, if you’re interested in working with me or checking out my other work, my links will be at the end of this email/post. And if you found value in this write-up, I would appreciate you sharing it with more people. It is word-of-mouth referrals like yours that help me grow. The best way to share testimonials is to share articles and tag me in your post so I can see/share it.