A few years into the new technology cycle, AI is reshaping the economy at a scale likely to exceed that of the last generation of software. What makes this shift so important is that it expands the ceiling on outcomes. AI is creating new willingness to pay by automating work itself and opening access to the $43T global labor spend. For Minnesota, the question is whether it will ride that generational economic wave or watch most of the value accrue elsewhere.

We looked at the data across startup funding, infrastructure investment, and the broader innovation stack. What emerged is a mixed but encouraging picture.

The Startup Picture:  Minnesota’s Early AI Cohort

Minnesota has real strengths in this cycle, and many of them are underappreciated. But foundation models are not one of them, and that should not be surprising. Companies like OpenAI and Anthropic are magnets for capital. They pull in elite talent, massive compute, and follow-on capital into a very small number of places.

Minnesota is not competing for that layer, and it likely should not try. The more important question is which layers of the AI economy it can realistically own and whether those strengths will translate into durable company creation and economic value inside the state.

In 2024 and 2025, Minnesota startups raised $2.8 billion (Pitchbook). Within the Midwest, that is an impressive number. Nationally, though, it represents just 0.5% of all startup funding raised over the same period. Minnesota’s ecosystem remains genuinely strong in a few areas, particularly healthcare and, to a lesser extent, fintech.Of the 103 Minnesota-based startups that raised at least $500,000 at Series A or earlier over the past two years, 22 are categorized as AI. An optimist could see the foundation of a real AI cohort in that. A different view would note that if AI is the defining startup cycle of this generation, 22 still looks relatively small.  Startup formation is only part of the answer, so now we need to look at whether Minnesota is capturing value elsewhere in the stack.

The Infrastructure Layer: Data Centers Are Coming

Large-scale data center development is coming to Minnesota. Meta is already building a major facility in Rosemount, Google announced a new project in Pine Island, and at least 12 large-scale data centers have now been proposed across the state (Minneapolis Fed).

Data centers are an important layer of the AI infrastructure because they drive demand for land, power, transmission, and physical compute infrastructure, all of which can reshape regional economies.

We will set aside the environmental controversy for now, but even on economic grounds, the picture is not especially clear about where the value accrues.  Hyperscalers tout job creation as a core local benefit, yet Brookings recently argued that data centers typically generate far fewer lasting jobs than many communities hope for.

A Forcing Function for Grid Innovation - The more interesting part of the Google-Xcel deal may not be the data center itself, but the grid buildout it is forcing in its wake. Alongside the Pine Island project, the two announced 1,900 MW of new clean energy resources, including a 300 MW Form Energy iron-air battery system that Xcel described as the largest battery project by gigawatt-hour capacity ever announced worldwide. In that sense, Minnesota is becoming a test case for how AI infrastructure can accelerate grid modernization.

Where Value May Accrue: Minnesota’s Industrial / Hardware Stack

This is where the narrative flips in Minnesota’s favor, and where most coverage misses the point. When people talk about the AI economy, they usually focus on model companies and software startups. But AI at scale also depends on a physical stack: semiconductors, storage, sensors, advanced materials, power electronics, and grid infrastructure. Minnesota is quietly deeper in several of those categories than it gets credit for.

Bloomington: Minnesota’s Quiet Hardware Corridor

Bloomington may be the most important place in Minnesota’s AI economy. It is home to Polar Semiconductor, SkyWater and Seagate. These two “foundries” anchor the state in semiconductor manufacturing, along with Seagate, which produces key components used in cloud storage infrastructure. Together, they show that Minnesota’s role in the AI economy may be stronger in the physical stack beneath large-scale compute than in the model or application layer.

Polar Semiconductor and Skywater are both “foundries.” A foundry is the manufacturing layer of the semiconductor industry. It takes chip designs and turns them into physical semiconductors. They both manufacture chips that end up in the real world, not inside data centers.  

Polar’s focus is on power, sensors, and high-voltage semiconductors, which are commonly used in industrial systems, medical devices, automotive applications, and aerospace and defense equipment.

SkyWater operates more like a specialty foundry. It manufactures chips for technically demanding, high-reliability applications, including defense, space, and other advanced systems where domestic production and process customization matter.

Seagate, while not headquartered in Minnesota, maintains a significant presence in the state and plays an important role in the storage layer of the AI infrastructure stack. The company builds mass-capacity storage hardware to meet the need for storing the enormous volumes of data that AI systems generate and retain.

Taken together, these companies put Minnesota on the map in chip manufacturing, storage hardware, and advanced electronics design. Their impact goes beyond the products they make. Over time, companies like these help train a regional workforce, anchor specialized expertise, and build the technical talent base that can support Minnesota succeed throughout this tech cycle.

Advanced Materials: The Old and the New

AI infrastructure depends on more than chips and servers. It also depends on the materials that help electronics manage heat, maintain reliable connections, improve power efficiency, and perform under demanding conditions.

3M represents the broader side of that story. Its materials science capabilities span data center infrastructure, including products for thermal management, connectivity, and other performance needs in modern computing systems.

Niron Magnetics is building the world's first full-scale manufacturing facility for rare-earth-free permanent magnets in Sartell, Minnesota, on the site of a former paper mill that closed over a decade ago. The technology uses iron and nitrogen instead of rare earth elements like neodymium, which China controls through roughly 90% of global mining and processing.

💡
Why This Matters for AI - Permanent magnets are inside data center cooling motors, EV drivetrains, industrial motors, and defense systems. China's control of rare earth supply has been a documented national security vulnerability. Niron's iron nitride technology, using Minnesota Iron Range ore, is a domestic solution to a critical supply chain problem.

The Invisible Layer of Value in Minnesota

Minnesota’s strengths are embedded in the physical infrastructure stack. That may be exactly what you would expect from a state whose economy has long been shaped by manufacturing, engineering, and materials innovation.

If you are thinking about building something, I would highly encourage you to take advantage of this tech transformation. It is an incredible opportunity, and we are still early!

Share this article

Share to Facebook
Share to X
Share to LinkedIn

Written by

Grant Gibson
Grant Gibson
Grant Gibson is a Principal at Great North Ventures, a Minneapolis-based Seed and Series A VC fund focused on Vertical AI and Fintech.