Most startup lawyers have never dealt with a Bayh-Dole march-in clause, flagged an ITAR deemed-export risk, or explained why a token sale in 2026 faces different SEC scrutiny than one in 2024. Deep tech founders, those building in synthetic biology, space, defense, crypto, AI, energy, nanotech, and robotics, need counsel who already understands the regulatory and IP terrain their companies operate in.
That’s what “deep tech” means to us, and it’s why we built a practice around it.
Imagine a synthetic biology founder spinning out of a university lab. She licenses the core IP from the university, raises a seed round on SAFEs, and starts building a pilot manufacturing facility.
Six months later, she needs counsel who understands how university license agreements interact with investor IP warranty requests, who knows that Bayh-Dole gives the federal government march-in rights if the technology isn’t adequately commercialized, and who can explain that her product might fall under FDA, USDA, and EPA jurisdiction at the same time.
Her BigLaw firm staffed a second-year associate with startup generalist experience. He had to research all of this from scratch, billing her $700 an hour.
Deep tech startups need deep tech counsel. They don’t need to be paying for generalist counsel to get up to speed.
Defining “deep tech”
The phrase gets used loosely, so it’s worth being precise about what we mean and why we at Altum Legal landed on this label.
Boston Consulting Group and Hello Tomorrow published the foundational research in 2017, defining deep tech as companies built around unique, protected, or hard-to-reproduce scientific and engineering advances.
Their data, drawn from 400+ startups, identified four defining challenges:
- Long development timelines
- High capital requirements
- Significant technology risk, and
- Markets that often don't exist yet.
A follow-up report in November 2024 put numbers on it: deep tech startups take 35% longer than conventional startups to grow from $1 million to $5 million in revenue. They consume 48% more funding to get there. Half don’t survive past seed.
MIT’s Regional Entrepreneurship Acceleration Program sharpened the definition in a 2023 paper. Their framework identifies four types of uncertainty that separate deep tech from conventional startups: technical risk, regulatory risk, market demand uncertainty, and scale-up uncertainty.
What’s in a name?
Deep tech startups and investors might use another phrase to define themselves. Perhaps they know the concept from Y Combinator, where the official directory tag is “hard tech.”
Some others have opted for “frontier tech,” but since 2023, “frontier AI” and “frontier model” have become established regulatory terms for the most capable AI systems. For a law firm advising AI startups, branding around "frontier tech" would mean constantly clarifying that we mean a practice area, not a regulatory category. We'd rather not explain the difference every time.
The legal challenges are identical regardless of which label you prefer. And as a Firm named for the Latin word for “deep,” perhaps it’s unsurprising we went with “deep tech.”
What makes deep tech different, legally
The defining feature of a deep tech company is that multiple systems of regulation, intellectual property, and corporate structure converge at the same time. The standard startup playbook doesn't account for these collisions.
A SaaS company raising a Series A deals with a cap table, employment agreements, commercial contracts, and maybe a privacy policy. Those are widely familiar problems with familiar templates.
A deep tech company raising the same round might deal with all of those, plus a university license that restricts how IP can be sublicensed, a government grant with specific compliance obligations, export controls that affect who you can hire, and a regulatory pathway that no agency has fully defined. No template library covers this.
Neither does a generalist lawyer’s pattern-matching from prior deals.
We work across seven verticals where these collisions are the norm: synthetic biology, space and defense, crypto, AI, energy, nanotech, and robotics. The regulatory details differ by sector, but the structural pattern is the same.
Three current examples show what that looks like in practice…
Synthetic Biology
Cell-cultured meat companies face a regulatory arrangement where the FDA oversees cell collection, growth, and differentiation, and then jurisdiction transfers to the USDA at the “harvest” stage. But cultivated seafood (except catfish) stays entirely under FDA. The pathway differs by species. And at least seven states have banned cultivated meat sales outright. On top of that, the USDA’s SECURE rule, which had modernized the framework for bioengineered crops, was vacated by a federal court in December 2024 in National Family Farm Coalition v. Vilsack; the rules governing your product can change backward, and recently did.
Space & Defense
Space and defense companies face a different version of the same problem. Most spacecraft and launch technology falls under ITAR. Hiring a foreign national engineer who accesses controlled technical data counts as a “deemed export" requiring a State Department license. Foreign investment triggers CFIUS review. And FAA Part 450, the commercial launch licensing framework that took effect in 2021, has turned into a bottleneck: the FAA had issued only eight Part 450 licenses as of mid-2025. SpaceX publicly accused the agency of delays costing millions per mission. And beyond the launch pad there’s jurisdictional confusion: No U.S. agency has clear statutory power to govern on-orbit commercial activities between launch and reentry.
Crypto
Crypto went through the fastest regulatory reversal of any sector. The SEC dismissed its enforcement actions against Coinbase (February 2025) and Binance (May 2025), and settled Ripple (August 2025) in rapid succession. Then, on March 17, 2026, the SEC and CFTC jointly issued a formal interpretive release establishing a five-category taxonomy for crypto assets. Four categories, including digital commodities, digital collectibles, digital tools, and payment stablecoins under the GENIUS Act, are explicitly not securities. Only tokenized traditional securities remain subject to federal securities laws.
A non-security crypto asset sold as part of an investment contract in a primary offering does not necessarily stay subject to that investment contract in secondary transactions. The agency went from suing the entire industry to publishing a formal classification framework in under two years.
But 49 states still require separate money transmitter licenses. Multi-state licensing runs $1 to $3 million over 12 to 18 months. So the federal picture got dramatically clearer in March. The state-level complexity didn’t change at all.
A generalist startup lawyer encountering any of these problems for the first time on your deal is learning on your dime, and certainly lacks the pattern-recognition that comes only with deep tech experience.
What about AI?
You’ll notice AI isn’t in the examples above. That’s because a pure-software AI startup raising a Series A looks a lot more like a SaaS company than a synthetic biology or space company. The cap table, the commercial contracts, the employment agreements all follow familiar patterns.
AI crosses into deep tech territory when it intersects with physical systems, regulated products, or dual-use applications.
An AI company building autonomous drone navigation falls under ITAR.
An AI company whose model is used in medical diagnostics faces FDA oversight.
An AI company deploying high-risk decision-making systems in Colorado must comply with the Colorado AI Act (SB 24-205), which takes effect June 30. And the EU AI Act's main enforcement provisions arrive August 2.
If your AI company touches any of those, the deep tech framing applies.
Deep tech is booming
Deep tech sometimes gets treated as a niche interest. The funding numbers say otherwise.
Synthetic biology attracted $12.2 billion in venture investment in 2024, per SynBioBeta's annual report. Defense tech hit a record $49.1 billion in 2025 according to PitchBook, with 10 new unicorns. Crypto ventures reached about $18.9 billion in 2025 per The Block, up 37% from the prior year. Space technology drew $26 billion in 2024, according to Space Capital.
Set pure AI aside. The remaining deep tech sectors combined represent roughly $50 to $70 billion in annual venture capital. That's 12 to 17% of all global VC, with each sector growing faster than the overall market.
The demand for lawyers who already know this terrain is growing apace with the capital.
What sector literacy actually means
Being sector-literate doesn’t mean being a regulatory specialist. It means three things.
Knowing which problems are genuinely novel and which are standard startup issues in unfamiliar packaging. A SAFE round is a SAFE round whether the company makes software or engineered organisms. But IP warranties in the term sheet mean something different when the IP originates from federally funded university research, where the government retains rights and the university holds the patents. Your lawyer should understand Bayh-Dole’s march-in provisions before the term sheet arrives.
Knowing when to handle something directly and when to bring in a specialist. We don’t do patent prosecution. We don’t do heavy regulatory work. But we flag ITAR exposure before the first foreign hire, not waiting until the State Department sends a letter. We coordinate with regulatory counsel before diligence reveals the problem, not after. The difference is whether the issue gets addressed in month three or discovered in month ten when a lead investor’s lawyer sends a request list and you can't produce clean answers.
Understanding the timeline. Deep tech companies take longer to reach revenue milestones and burn more capital getting there. The legal relationship runs longer than a typical startup engagement. The cap table gets more complex earlier, with government grants, university licenses, and strategic investors layered on top of standard venture terms. A missing IP assignment that would cost $5,000 to fix at formation can cost $50,000 to unwind during Series A diligence, not to mention negotiation leverage.
A quick test
If your company involves any two of the following, you’re in deep tech territory:
- IP that originates from government-funded or university research
- Overlapping regulatory jurisdictions, or a product category without a settled regulatory framework
- Physical products, biological materials, or hardware requiring agency approval
- Export controls or dual-use technology
- A development timeline measured in years, with capital needs to match
Two or more, and your lawyer should already know what those things mean when you walk in the door. The learning curve is steep, the regulatory landscape moved significantly in the last twelve months alone, and you shouldn't be paying hourly for someone to catch up.