Deep Tech Brief #03

Welcome to this week's issue of Deep Tech Brief.

Every Saturday, we break down deep tech companies building genuinely hard things, in plain language. No jargon, no hype.

Three companies this week, and they look unrelated at first: a Slovenian team storing data inside DNA, a German startup importing Chinese electric trucks, and a Nordic team that turns a typed sentence into working hardware.

The thread is that none of them is really betting on inventing the core technology. In each case the hard science, or the manufacturing scale, already exists somewhere.

The bet is on access: taking a frontier capability locked behind a lab, a border, or a specialist, and turning it into something an ordinary buyer can use.

The world is producing data faster than it can store it. Global data volume is expected to pass 175 zettabytes and climb toward roughly 380 zettabytes by 2028, and by most estimates less than 2% of it will ever be preserved. The media we rely on do not help: hard drives and magnetic tape are power hungry, last only a decade or two, and need constant migration. Analysts already forecast a storage supply gap later this decade. That is the problem DATANA is aiming at.

DATANA, a project spun out of the Slovenian biotech firm BioSistemika in Ljubljana, encodes digital files into synthetic DNA, the same four-letter molecule nature uses to carry genetic information. The pitch is easy to feel even if the chemistry is not. By Luka Zupancic's account, a single small vial holds about 200 petabytes, roughly the contents of an average data center. Industry figures put DNA's theoretical ceiling near 215 petabytes per gram, so the claim is in the right universe. Once written, the DNA needs no power, stays stable for tens of thousands of years at room temperature, and is inherently immutable, quantum-resistant, and EMP-proof.

The clever part is not the density, which every DNA storage group can claim. It is that DATANA is actually selling. Most names in this field (Twist Bioscience, Catalog, France's Biomemory) are still in the lab or shipping novelty consumer kits. DATANA launched a commercial service this year, Molecular Wallet, aimed at ultra-secure storage for cybersecurity and digital-asset customers. In a pilot with the crypto exchange Bitstamp, the company says it stored and retrieved a full crypto wallet backup, keys, seed phrase, and address, in DNA, which it bills as a world first.

Here is the honest limit. The bottleneck for the whole field, DATANA included, is throughput: how fast and how cheaply you can write DNA. To see the gap, consider that the largest published DNA archive to date, around 200 megabytes, needed nine separate synthesis runs, and synthesis has historically cost well over $100,000 per gigabyte. Luka names throughput as the central challenge, and the capital raise now underway is meant to push encoding capacity up.

Worth watching, because the sustainability and security case is genuinely strong and DATANA is further toward market than most rivals. But the economics of writing DNA at scale remain unproven, and that, not the science, is what will decide whether this becomes infrastructure or stays a high-end vault for tiny, priceless files.

Learn more: https://datana-storage.com

Heavy trucking is one of the hardest corners of transport to electrify, and Europe is behind. Battery-electric trucks were just 4.2% of EU heavy-duty sales in 2025, up from 2.3% the year before, while in China zero-emission trucks are already about 29% of the market. The blocker is price. A European electric heavy truck averages around 320,000 euros against roughly 100,000 for a diesel. Meanwhile EU rules demand deep cuts in truck emissions, 45% by 2030 rising to 90% by 2040, so fleets have to electrify whether the math is comfortable or not.

Kraftvolt, founded by Holger Dornieden, wants to close that gap by importing Chinese electric trucks into the German market. By Holger's account, the Chinese manufacturer he works with built its expertise through a joint venture with Mercedes-Benz, learned the craft, and is now entering Germany on its own. Whether or not that origin story holds for the specific partner, the broader trend is real: more than six Chinese truckmakers, including BYD, Sany, Sinotruk and Windrose, are moving into Europe in 2026, some priced up to 30% below European equivalents.

Kraftvolt's product is not the truck. It is the bridge. Holger's edge, by his own telling, is 25 years of working with Chinese firms and the ability to navigate a cross-cultural, certification-heavy market that trips up newcomers. That matters, because the Chinese advantage is not just cost but speed: the startup Windrose developed and certified a heavy electric truck in about three years, against a typical seven-year industry cycle.

One to watch, carefully. The tailwind is undeniable and the timing is sharp. But a relationship is a thin moat. Kraftvolt is one of several would-be intermediaries, and the biggest Chinese brands are already building their own European sales networks, some setting up assembly inside the EU. Holger's stated goal is a 100 million dollar business, and the honest test is whether he can lock in supply and after-sales support before the manufacturers route around the middleman.

Learn more: https://kraftvolt.gmbh

Software got easy. A teenager can ship a web app in a weekend using layers of tools that hide the hard parts. Hardware never got that treatment. Building a custom device still means schematics, component selection, firmware, and manufacturing, and by Martin Yang Eriksson's account, even an experienced engineer using an Arduino or Raspberry Pi should expect two to three weeks to get a working prototype. That gap is what Atech is attacking.

Atech, based across Copenhagen and Stockholm and founded in 2026, calls itself "Lovable for hardware." You describe what you want in plain language, "a device that tracks the heat in this room" or "a console that plays Snake," and the platform picks compatible modules from a kit, wires up the configuration, and generates the firmware. You snap the parts together and upload the code, in minutes rather than weeks. The company calls it "vibe-engineering," and its bet sits inside the fast-rising "Physical AI" thesis that Nvidia and others have flagged as a major next market: systems that sense and act in the real world.

One correction worth making, because the numbers matter. Atech raised a pre-seed of about 800,000 dollars, led by Nordic Makers and Emblem, with participation from Lovable and an angel check from Lovable's CEO. Sequoia and Andreessen Horowitz are involved, but through their Scout funds, not their flagship funds, a smaller and more speculative signal than "backed by Sequoia and a16z" implies. The team, led by CEO Tomas Erik Harmer with Martin as Head of Go-to-Market, is early and small.

One to watch. The problem is real and the demo genuinely lands. The open question is how far the abstraction reaches. Snapping together a curated kit is one thing; arbitrary hardware, where a wrong choice on a circuit board or a firmware timing bug becomes a dead physical object rather than a quick code fix, is another. If Atech can stretch from modular kits toward truly open-ended hardware, it is building something that does not exist today.

Learn more: https://atech.dev

The through-line: in all three cases the frontier already exists, in a lab, a Chinese factory, or an engineering discipline, and the company's real product is the on-ramp that lets the rest of us reach it.

Know a deep tech company we should break down next? Hit reply and tell us. The best editions come from real tips.

See you next Saturday.