Deep Tech Investing: Moving Beyond MVPs with Minimum Valuable Probes

Why Deep Tech Is Suddenly Everyone’s Business

Let’s talk about deep tech. It’s not your typical startup scene. This is where science and serious engineering collide to build the stuff that could genuinely change the world—like quantum computers, next-gen energy solutions, or life-saving biotech. These companies aren’t running on quick wins or off-the-shelf software. They need years of R&D, expensive infrastructure, and often a tight connection with universities, labs, and industry partners.

Lately, deep tech has been catching serious attention from investors. A recent report from the Boston Consulting Group showed that in 2023, deep tech made up about 20% of all venture capital globally—twice what it was ten years ago. Europe’s leaning in even harder. Nearly half (44%) of all VC funding on the continent last year went into deep tech projects. That’s not just a trend—it’s a shift in priorities.

And it’s not just about money. There’s a bigger play happening here. The EU recently rolled out a massive €20 billion plan to build out artificial intelligence “gigafactories”—basically, their way of showing they’re ready to compete with the US and China in the tech arms race. At the same time, the European Innovation Council has already committed more than €7 billion to a range of deep tech startups focused on solving massive challenges in energy, health, space, and advanced computing.

This focus isn’t just regional either. France saw deep tech funding shoots up by 38% last year—even though startup investment overall was dropping. In the UK, leaders are pushing for a similar ambition into deep tech to avoid losing top talent and stay competitive on the global stage.

But here’s the catch: deep tech doesn’t play by the usual startup rules. The whole “build fast, test faster” playbook? It tends to fall flat when you're working with things like nuclear fusion or synthetic biology. These aren't apps you can tweak overnight based on early user feedback. They’re complex, slow-moving, and full of unknowns.

That’s where a different approach comes in—something called the Minimum Valuable Probe, or MVP (yep, a different kind of MVP). Instead of building a half-baked product and tossing it out there, a Minimum Valuable Probe is more like a focused experiment. It's meant to test out one or more big assumptions—like, is this technically possible? Will people use it? Is it economically viable? Can it scale? Can it be done sustainably?

These probes aren’t products. They might look like a simulation, a research partnership, a test with a potential customer, or even a conversation with regulators. The goal is to learn fast but in a way that makes sense for deep tech—where the stakes are high, and the timelines are long.

So, as we dive into this idea throughout the rest of the article, just keep this in mind: Deep tech isn’t about moving fast and breaking things. It’s about moving wisely and proving things. That’s where MVPs—Minimum Valuable Probes—come into play.

Why Investing in Deep Tech Feels Like Playing Chess in a Maze

Backing a deep tech startup isn’t like throwing money at the next photo-sharing app or food delivery service. These ventures sit right where big science meets bold business, and that combo brings its own unique headaches—especially for investors trying to pick winners.

Unlike most traditional startups that build on familiar tools or tweak existing products, deep tech companies often revolve around brand-new science. We're talking tech that hasn’t even proven itself yet. Add in long timelines, big budgets, and a whole lot of “what ifs,” and you’ve got a different ballgame entirely.

First Big Hurdle: Will It Work, Then Will Anyone Want It?

Most venture capital bets boil down to one question: Will people buy this? But in deep tech, investors don’t even get to ask that until after answering something way harder: Can this even be built?

These startups often need years—sometimes decades—just to prove the tech itself works. That’s years of research, testing, and navigating regulations before you even get a chance to ask if anyone’s interested in buying. And even then, because the tech is so new, it’s often unclear who the customer is or whether a market even exists.

So, deep tech investors have to think like both scientists and business strategists. Funding isn’t just about throwing gas on a fire—it’s more like funding a series of experiments. Each check should unlock some new insight, not just another version of the product.

Time and Money: You’ll Need a Lot of Both

In the software world, a startup can push updates every week. Deep tech? Not so much.

If you're building something like a quantum computer, a nuclear fusion reactor, or a new form of biotech, you’re in for a long haul. That means expensive labs, physical prototypes, slow feedback loops, and endless trial and error.

And here’s the kicker—most VC funds are structured to last 10 years. But many of these technologies won’t be ready for prime time until year 12 or 15. Some investors are getting creative—launching evergreen funds or teaming up with governments and corporations to buy more time.

The Dreaded “Valley of Death” Hits Harder in Deep Tech

There’s a stretch in every startup’s life where things get dicey—that moment between proving the idea works and making it actually sells. In deep tech, that stretch is longer and scarier.

Founders might scrape together early funding to show their science is solid, but once they need serious money to build a business around it, investors start getting skittish. Venture capitalists might think it's too risky. Big corporations might think it's too early. So the startup gets stuck in limbo.

The only way out? Strategic moves like pilot programs, public contracts, or partnerships with bigger players—basically anything that helps prove there's a real-world path forward.

You’re Not Just Selling a Product—You’re Orchestrating an Ecosystem

A lot of deep tech companies can’t just launch and go. They need to win over regulators, scientists, manufacturers, end-users, sometimes even lawmakers. Think of a new healthcare technology—it needs to play nice with clinical trials, government approvals, health insurers, and patients. It’s not just about go-to-market—it’s go-to-system.

That’s why smart investors look for signs that the startup can play nice with the whole ecosystem. Early partnerships, pilot runs, or even small government grants can help show the company’s ready to operate in the real world—not just in the lab. An ecosystem can refer to the internal conditions favouring or hampering the flow of ideas (or hypotheses) making almost impossible any change or innovation. Here the roots are quite similar to the theory put forward by Edmonson about the role of "psychological safety" in organizations. On the other hand you have also the ecosystem referring to the socio technical set up characterising the markets surrounding your business model. I am afraid that in this case both are critical for our investments.

What This Means for Investors

If you’re a deep tech investor, you’re not just writing checks. You’re helping shape an entire learning journey.

Your role isn’t just about scaling up companies—it’s about helping them figure out what works and what doesn’t. It's about sensing patterns, running smart experiments, and encouraging founders to engage with reality rather than just polish their pitch decks.

That’s where Minimum Valuable Probes (MVPs) come in. They’re tiny tests that help founders and investors alike learn something important, fast. Not every experiment needs to lead to a product launch. Sometimes, the biggest win is just learning what not to do next.

This mindset shift—from scaling to sensing, from plans to patterns—is what turns deep tech investing from a gamble into something truly transformative.

Minimum Valuable Probes: A Smarter Way to Test Ideas in Deep Tech

We’ve all heard of the MVP—the Minimum Viable Product. Eric Ries, the guy who coined it, described it as the simplest version of a product that helps you learn the most about your customer with the least effort. But it’s more than just a half-finished app or bare-bones prototype. An MVP is meant to be an experiment—a way to test if you’re on the right track.

Now, here’s the thing. In deep tech—where you’re dealing with long development cycles, tricky regulations, and emerging markets—it’s not always possible or even useful to build a quick, usable product. And that’s where people get tripped up. They try to apply the MVP playbook from software startups to something like quantum computing or next-gen materials, and it just doesn’t work.

So what do you do instead?

Say Hello to the Minimum Valuable Probe (MVProbe)

Think of an MVProbe as MVP’s deep-tech-savvy cousin. It’s a small, strategic experiment designed not to ship something, but to learn something critical—specifically, across five key areas: value, usability, feasibility, viability, and sustainability.

You’re not trying to launch a product. You’re trying to get answers.

For example, if you’re working on quantum tech, your “product” might not be ready for years. But you can build and test photon sources, run simulations, or partner with research labs to see if your approach is even possible. That’s what Quandela did, and it worked—they learned what they needed without rushing a market-ready device.

So, What Makes MVProbes So Useful?

In deep tech, you're often building something the world hasn’t seen before. That means your biggest challenge isn’t just building—it’s figuring out what needs to be true for the thing to ever work out.

Trying to build an actual product too early wastes time and money. What you need are fast, focused ways to test whether your key assumptions hold up in the real world. MVProbes give you that. They help you answer questions like:

• Can we build this at scale?

• Will customers—or regulators—buy in?

• Is this something people can realistically use?

• Will it ever make money?

• Should it even be built, from a sustainability or ethics angle?

And here’s the best part: MVProbes aren’t locked into one format. They’re flexible. If you're testing feasibility, you might build a simulation or run a bench test. To explore value, maybe you have in-depth interviews or mock up a partnership offer. Testing usability? Try a pilot run in a controlled environment. Worried about regulatory issues? Set up a roundtable or policy experiment.

MVProbes Help You Navigate the Unknown

Instead of committing to a single solution too early, you use MVProbes like flashlights. They help you light up dark corners of your roadmap before you make expensive moves. They’re like strategic bets—you don’t know if they’ll pan out, but each one gives you valuable intel.

Think of it this way: MVPs tend to ask, “Do people want this?” MVProbes ask, “What has to be true for this to succeed in the real world?” And that subtle shift makes all the difference in complex, long-game innovation.

In deep tech, insights are sometimes more valuable than early traction. One sharp realization can save you months of dead-end work.

How to Use MVProbes to Manage Risk in Deep Tech

Deep tech isn’t like building an app or running a DTC brand. The risk landscape is way more layered. It’s not just about product-market fit. You’ve got science, supply chains, regulations, ethics, infrastructure—you name it.

To stay sane (and smart), you want to break risk down into five big buckets. Each one maps to a kind of MVProbe you can use to learn early and adjust course.

Feasibility Risk

Can we even build this? Just because something works in a lab doesn’t mean it’ll hold up in the wild. Before scaling, run simulations or stress tests. See if it survives the jump to real-world conditions.

Value Risk

Will anyone care? You might think your tech is amazing, but if it doesn’t solve a felt need, it’ll flop. Use interviews, mock sales conversations, or partnerships to get clarity on whether users want what you’re cooking.

Viability Risk

Is this a real business? Even with good tech and some interest, your business model might not hold water. Can you price it right? Will regulators sign off? What are your customer acquisition costs? Run financial experiments, test pricing, or sketch out different go-to-market paths.

Usability Risk

Can people figure out how to use this thing? Whether it’s surgeons, farmers, or factory workers, someone’s got to use your tech. That means integration, UX, and workflows matter—sometimes more than you’d expect. Pilot tests or roleplay environments help here.

Sustainability Risk

Should we even build this? This one’s getting more important by the day. Does your innovation play nice with the planet? Will future policies embrace or ban it? Is it ESG-ready? Probes like life cycle assessments or policy simulations can help you spot red flags before regulators or the public do.

Real-World Examples: How Companies Used Small Experiments to Reduce Big Risks

Before going all-in on a new product, smart companies often run small, strategic tests to make sure they’re on the right track. These tests—sometimes field trials, simulations, limited releases, or demos—help spot potential problems early and prove there's real value before making huge investments. Let’s look at how a few innovative companies used this approach to their advantage.

Pivot Bio – Rethinking Fertilizer with Microbes

Pivot Bio set out to change how farmers fertilize crops. Instead of relying on synthetic nitrogen, they created a microbial solution that helps plants feed themselves. But rather than rushing their product to market, they ran tons of field tests across different types of farms and crops.

These early trials, kind of like “mini experiments,” let them see how well the product worked in real-world conditions. They collected their own performance data—something they called YODA—which showed consistent improvements in crop yields. This gave them solid proof their idea worked and helped win over regulators and farmers. Taking this careful, step-by-step path made sure their product delivered real value before going big.

TerraPower – Reinventing Nuclear Energy

Backed by Bill Gates and a group of tech investors, TerraPower is working on advanced nuclear reactors. As you can imagine, building a nuclear plant isn’t exactly something you do overnight. It’s expensive, super complex, and under heavy regulation.

So, instead of building a prototype right away, they started with computer models and detailed safety reports. These became their version of “minimum valuable probes”—tools to test if the idea could work and to build confidence with government agencies like the Department of Energy.

They used each phase of development not just to improve the tech, but also to get buy-in from key partners. For deep tech startups like TerraPower, these kinds of stakeholder-focused MVPs can be just as important as the tech itself.

Quandela – Building Quantum Tech One Piece at a Time

Quantum computing is incredibly powerful—but it’s also tricky to build. Quandela didn’t try to create a full quantum computer right away. Instead, they started with one small but essential piece: a reliable single-photon source.

This “start small” mindset let them work out technical bugs early while proving the value of their tech. They teamed up with big players like EDF to simulate real-world problems, such as the stress on hydroelectric dams. Later, they delivered a quantum computer to OVHcloud to test how well it worked in a cloud environment.

Each project helped answer a different question—Can we build it? Will it be useful? Is it scalable? In the end, these steps made their platform stronger and more trusted, especially after getting selected by EuroHPC for major European quantum initiatives.

Lytro – A Lesson in What Not to Do

Lytro had some seriously cool technology—a camera that could refocus images after they were taken. But having cutting-edge tech doesn’t guarantee success.

Their first product worked from a technical standpoint, but it didn’t play nicely with existing software and was confusing for everyday users. Later versions aimed at professionals looked promising but didn’t fit into standard workflows. Everything felt a bit too complicated and out of sync with what people needed.

Lytro’s story is a reminder that even brilliant innovations can flop if they don’t align with how users think, work, and create. MVPs shouldn’t just test if something works—they should also make sure people want to use it and can use it without hassle.

MVP-Led Startups vs Traditional VC-Funded Ones: What Really Works in Deep Tech?

When it comes to launching deep tech startups—think quantum computing, nuclear power, or biotech—there are generally two ways founders and investors go about it. One is the old-school, build-first, raise-big-money approach. The other is a newer, more thoughtful strategy built around testing and learning early, often through what’s called Minimum Valuable Probes (MVPs). So, how do these two paths play out?

MVP-Led Startups: Learn First, Build Smart

In MVP-led startups, the focus isn’t just on building a product fast. It’s on learning quickly and reducing uncertainty as early as possible. Founders and investors work together to run small, smart experiments that uncover key answers: Will regulators support this? Will customers care? Can this work in the real world?

These MVPs—whether they’re field tests, simulations, or pilot partnerships—help the team make informed decisions. They prioritize what needs to be proven first instead of rushing into full production. It’s about solving the right problems in the right order, with just enough investment at each step.

Traditional VC-Funded Startups: Build Now, Worry Later

On the flip side, many traditionally funded deep tech startups dive headfirst into building their product based mostly on scientific breakthroughs. They put a lot of weight on technical validation and less on commercial testing or real-world usability.

This approach can work in certain research-heavy fields, but it has its blind spots. When companies delay testing real-world fit—like how users interact with the product or whether it fits into existing ecosystems—they risk hitting major roadblocks down the line. Sometimes the product works beautifully in the lab but flops in the market.

Case in Point: Lytro vs. Quandela

Take Lytro, for example. Their light-field camera was a technical marvel, but it didn’t click with users. The product needed proprietary software and didn’t fit into photographers’ normal workflows. So even though the tech was solid, the market just wasn’t ready—or maybe it was the wrong market altogether.

Now compare that to Quandela, a quantum startup that took the MVP-led route. They didn’t try to build a complete quantum computer out of the gate. Instead, they rolled out smaller components, like single-photon sources, and tested each piece with strategic partners. Every step was designed to answer a specific risk: Can we build it? Will it add value? Will people use it?

By working through these layers one at a time, Quandela built trust, gathered proof, and lined up key support—all before scaling up.

The Big Takeaway

The real difference between these two paths comes down to mindset. Traditional approaches often ask, “What can we build?” MVP-led teams ask, “What do we need to learn next?”

That simple shift changes everything. It turns product development into a discovery process. And in deep tech—where the unknowns run deep and the stakes are high—learning early can save years of effort and millions of dollars.

So, here’s the bottom line: Success in deep tech isn’t just about inventing something amazing. It’s about aligning that innovation with real needs, proving it step by step, and building a clear path from idea to impact.

Sources

Sources

Boston Consulting Group

“Deep Tech: The Great Wave of Innovation” Source of data on global deep tech VC funding and investment growth trends. https://www.bcg.com/publications/2023/deep-tech-wave-of-innovation

Lakestar 2025 European Deep Tech Report

Insight into Europe's funding resilience and strategic shift. https://lakestar.com/news/2025-european-deeptech-report

European Innovation Council (EIC)

€7B deep tech funding portfolio and strategic goals. https://eic.ec.europa.eu/news/european-innovation-council-impact-report-2023-eu70-billion-deep-tech-portfolio-2024-03-18_en

The Guardian

“EU to build AI gigafactories in €20bn push to catch up with US and China” https://www.theguardian.com/technology/2025/apr/09/eu-to-build-ai-gigafactories-20bn-push-catch-up-us-china

Le Monde (March 2024)

• Deep tech investment in France increased by 38% year over year. https://www.lemonde.fr/economie/article/2024/03/22/les-deep-tech-un-pari-qui-reste-a-consolider-pour-la-french-tech_6223517_3234.html

The Times (UK)

• Strategic urgency around deep tech to prevent brain drain. https://www.thetimes.co.uk/article/investors-offered-training-in-understanding-cutting-edge-tech-enterprise-network-thtbzc5cw

🧪 Case Studies

Quandela (Quantum Photonics)

Photon source development: https://www.quandela.com/resources/blog/entangled-photon-generators-for-fault-tolerant-photonic-quantum-computers

EDF partnership for simulation: https://www.quandela.com/about-us/newsroom/quandela-and-edf-work-together-to-use-photonic-quantum-computing-to-simulate-hydroelectric-dam-structures

OVHcloud delivery: https://sifted.eu/articles/french-startup-quandela-raises-e50m

EuroHPC contract: https://www.cea.fr/english/Pages/News/quandela-attocube-systems-AG-photonic-quantum-computer.aspx

Lytro (Light Field Imaging)

Product history: https://www.wired.com/2012/10/lytro-camera-light-field/

Lytro Cinema announcement: https://techcrunch.com/2016/04/11/lytro-immerge-gets-immersive/

What went wrong: https://spectrum.ieee.org/what-really-killed-lytro

Framework References

Eric Ries – The Lean Startup Definition and application of the MVP concept as an experiment for validated learning.

Alberto Savoia – The Right It Pretotyping, YODA (Your Own DAta), and how to test desirability quickly.

MIT Disciplined Entrepreneurship (Bill Aulet) 24-step methodology for de-risking venture design.