Deep Tech Briefing #29: 🌊 Underwater Data Centers in San Francisco 🚜 Kubota Acquires AI-Agritech; 🚀 UAS Defense Startup Secures $21M; ⚛️ Quantum Computing Breaks New Record and more...
An insider’s update on Deep Tech Ventures: Your dose of tech innovations, startups, exponential industries, policies, and market moves to stay ahead and capitalize on it.
Hey there! Welcome to this edition of Deep Tech Briefing, our Sunday column where we break down the week’s top developments in Deep Tech Startups and Venture Capital.
Subscribe for free to get it straight to your inbox every Sunday, and read it whenever you like!
In today’s edition
Underwater data centers test in San Francisco Bay for energy efficiency.
Kubota buys AI-agritech startup, boosting precision agriculture.
Defense startup Darkhive raises $21M for secure UAS solutions.
NanoSyringes advances drug delivery with a £10 million investment.
Quantum computing achieves a record 99.9993% SPAM fidelity.
✨ For more, see Membership | Partnership | Deep Tech Catalyst
🔸Underwater Data Centers in San Francisco Bay: “Move Fast and Ask for Forgiveness” Strategy for a Freshly Graduated from YC?
NetworkOcean, freshly graduated from Y Combinator, has unveiled its bold plan to test an underwater data center in San Francisco Bay. According to Wired, however, there’s a glaring issue: the company has not yet sought permission from the relevant authorities to proceed with its test.
Data centers, notoriously energy-hungry, present a significant challenge for environmental policies, with some major operators consuming more energy than entire cities. In fact, projections indicate that by 2030, data centers could be consuming a staggering 1 trillion gallons of water annually. By submerging its data centers in the ocean, NetworkOcean claims it can reduce water consumption to zero and cut energy use by 25%, leveraging the natural cooling properties of seawater.
Their underwater capsule, equipped with 2,048 Nvidia H100 GPUs, signals their intent to handle high-performance artificial intelligence and machine learning workloads. This also promises significant cost savings and reduced latency for edge computing, particularly in coastal areas like San Francisco.
Although Microsoft experimented with underwater data centers through its Natick project, NetworkOcean’s endeavor is on a completely different scale and in a unique environment like San Francisco Bay.
Regulatory hurdles in the Bay Area are notoriously stringent, particularly when it comes to environmental protections. San Francisco Bay is a highly regulated ecosystem, and any project potentially affecting it—especially one involving industrial equipment submerged in the water—faces intense scrutiny. Following Wired's inquiries, at least two critical agencies, the Bay Conservation and Development Commission (BCDC) and the San Francisco Regional Water Quality Control Board, have already issued letters warning NetworkOcean that proceeding without permits could result in significant fines.
Obviously, this regulatory gap raises concerns about how prepared NetworkOcean is to navigate the complex legal and environmental landscape surrounding the project. Securing approval from these agencies isn’t just a formality—it’s a lengthy and arduous process, often requiring extensive environmental impact assessments, public consultations, and negotiations with multiple stakeholders.
Nonetheless, the allure of reducing energy consumption and enhancing efficiency through natural cooling is undeniable. Data centers, as energy-demanding infrastructures, remain a major challenge for environmental policies, with some operators using more energy than entire cities. In this context, NetworkOcean could hold a winning card if they can prove that their project indeed has a smaller environmental footprint compared to traditional data centers.
The recent push to deploy an underwater data center in San Francisco Bay underscores a growing trend of pushing technological boundaries to solve critical infrastructure challenges. At the same time, it highlights the friction between innovation and regulatory frameworks. It also presents a direct challenge to policymakers: How do you foster the innovation required for next-gen data infrastructure while safeguarding ecological systems? Moreover, how will startups, often moving faster than regulatory frameworks, navigate the complexities of obtaining approval in sensitive ecosystems?
The upside for stakeholders? If the environmental concerns can be successfully managed, the potential to deploy high-density computing solutions without the traditional energy and water requirements of land-based data centers is game-changing. Underwater data centers could serve as a model for decarbonizing the digital infrastructure—a high-impact area that will only grow as demand for cloud services and AI accelerates.
🔸 Kubota Corporation Acquires Pittsburgh-Based AI Startup, Aiming to Transform Agriculture
In a strategic move that underscores the growing intersection of artificial intelligence and agriculture, Kubota Corporation has acquired Pittsburgh-based startup Bloomfield Robotics. This acquisition highlights important market dynamics and strategic alignments within the agritech landscape. Bloomfield, recognized for its advanced crop monitoring technology that utilizes AI and high-resolution imaging, represents a pivotal step for Kubota as it seeks to establish a dominant position in the precision agriculture sector.
For both companies, this move marks a significant milestone. Kubota solidifies its commitment to digital transformation in agriculture, while Bloomfield—a company founded just four years ago, emerging from early research at Carnegie Mellon University’s Robotics Institute—enters the global industrial stage with a heavyweight partner.
While the financial details of the transaction remain undisclosed, the acquisition aligns seamlessly with Kubota’s broader push toward digital agriculture. This positions Kubota to better compete with industry giants such as John Deere, which has heavily invested in autonomous tractors and AI-driven agricultural solutions. With the global market for precision agriculture projected to reach $12 billion by 2025, according to Markets and Markets, integrating Bloomfield’s technology could provide Kubota with a critical edge in this rapidly evolving market.
Specifically, Bloomfield's focus on vineyards and orchards offers Kubota an opportunity to target niche segments with tailored solutions, providing a potential competitive advantage over more generalized approaches adopted by its competitors.
Looking at the broader context, this deal comes at a time when the agricultural sector is facing unprecedented challenges. From the impacts of climate change to the rising global demand for food, the pressure on companies to enhance efficiency and sustainability has never been greater. In this environment, AI is emerging as a key enabler. Although Bloomfield is a smaller player, its technology could significantly reshape modern farming by facilitating more precise, data-driven decision-making that enhances yields and reduces waste.
However, there are hurdles ahead. A key question is whether Bloomfield can scale effectively within Kubota’s operations, and whether this acquisition will translate into market-leading innovations or become an underutilized asset with high potential. The success of this integration will depend heavily on Kubota’s strategic execution and its ability to leverage Bloomfield's AI capabilities across its product lines, driving measurable improvements in agricultural productivity and efficiency.
Ultimately, on a broader industry scale, this acquisition could serve as a bellwether for the industry, highlighting the increasing value placed on specialized, AI-enhanced agritech solutions and prompting other companies to pursue similar strategic deals as a means of fostering innovation and expanding market reach —a move that could alter competitive dynamics and investment trends across the agritech sector.
🔸San Antonio, Texas Defense Startup Raises $21M in Series A for Advanced UAS Security Solutions
The world of defense technology is changing rapidly. As the need for secure and adaptable solutions grows, Uncrewed Aerial Systems (UAS) are emerging as key players in this transformation. With the increasing importance of protecting critical infrastructures and enhancing national security, investments in this space are surging.
In this evolving landscape, innovative companies are drawing the attention of venture capital firms, particularly those with an eye on cybersecurity and national defense. One such company is Darkhive, a promising defense startup based in San Antonio. Recently, Darkhive raised $21 million in a Series A funding round. The round was led by Ten Eleven Ventures and included contributions from Crosslink Capital Inc., RTX Ventures, and Stellar Ventures. This influx of capital is a testament to the growing recognition of the value secure UAS solutions bring to the defense industry.
Darkhive has carved out a niche by developing lightweight, advanced drones tailored for short- and medium-range reconnaissance missions. At the heart of the company’s offering is a proprietary software platform that is not only secure but accredited by the Department of Defense (DoD). This platform allows for continuous software updates and provides comprehensive fleet management—features that are crucial in high-pressure environments where military personnel rely on real-time data to make mission-critical decisions. In these scenarios, managing a fleet of drones securely and efficiently can mean the difference between success and failure.
Accordin to John Goodson, CEO of Darkhive, a significant challenge in the UAS market is the widespread use of Chinese-made technologies. While these systems are often affordable and easy to use, they come with serious cybersecurity risks. This puts operators in a difficult position: choosing between insecure but cost-effective solutions and secure options that can be expensive and complex to deploy. Darkhive aims to solve this problem by offering a UAS architecture that seamlessly blends security with simplicity and operational flexibility.
The competition in the UAS space is heating up. Companies like Skydio raised $230 million in 2023 to develop AI-powered drones, showing how much value is being placed on automation and machine learning in modern defense strategies. But the real game-changer in this industry is the ability to deliver robust security without compromising efficiency. As defense budgets increasingly prioritize cybersecurity, companies that integrate secure software ecosystems into their UAS platforms are positioned to lead the market.
It’s also worth noting that UAS platforms, while vital in military settings, have uses that extend far beyond the battlefield. They’re being deployed in emergency response, infrastructure monitoring, and more. This versatility has sparked significant investment in UAS research and development worldwide, as governments and businesses race to enhance the security and reliability of these systems.
As the defense tech sector continues to evolve, the role of cybersecurity in defense is becoming impossible to overlook. A recent report projects the global cybersecurity market will skyrocket to $376 billion by 2029, with defense playing a major role in that growth. As venture capital flows into this space, the focus is shifting toward solutions that not only provide cutting-edge technology but also strong cyber resilience.
🔸Advanced Intracellular Drug Delivery Tech Secures Multi-Million Pound Funding
Intracellular delivery is often dubbed the “Holy Grail” of drug delivery for a reason. Traditional methods, such as viral vectors or liposomes, suffer from inefficiencies, off-target effects, and potential safety concerns. These limitations severely restrict the therapeutic potential of biologics, especially for treating diseases like cancer, genetic disorders, and autoimmune diseases, which often require precise targeting within the cell.
NanoSyrinx has developed a technology that addresses this issue with a completely new approach: "nanosyringes." This week, it announced the closing of a £10 million funding round aimed at advancing its technology for the targeted intracellular delivery of biological therapies.
So, what makes this important? The nanosyringe is constructed using a genetically encoded system that self-assembles during a bacterial fermentation process, making production scalable and cost-effective. Once assembled, these nanosyringes are loaded with therapeutic molecules and guided by cell-targeting arms that ensure high specificity, significantly reducing the off-target effects typically seen with traditional delivery methods.
What’s more, direct delivery of therapeutic agents into the cytosol bypasses cellular degradation pathways such as the endosome-lysosome system. This ensures that the biological payload reaches its target without being prematurely broken down, resulting in better treatment efficacy. This advancement opens up the possibility of addressing previously “undruggable” intracellular targets, such as mutated proteins in cancer cells (for instance, the RAS mutation).
Of course, like any new therapeutic delivery platform, there are regulatory hurdles to consider. The nanosyringe technology will need to clear rigorous clinical trials to prove not only its efficacy but also its safety. Given that the technology mimics bacterial systems, there may be concerns around immunogenicity or unwanted side effects. However, the recent surge of interest in biologics and gene therapies suggests that regulators are increasingly familiar with complex biological systems, which could work in favor of this new platform.
The recent funding round, led by BGF, Octopus Ventures, M Ventures, and supported by Eli Lilly, signals strong confidence in the technology. This capital infusion, coupled with the appointment of Dr. Edwin Moses, a proven leader with a track record of success in the biotech sector, strengthens the company's position as it expands operations and moves toward commercialization.
Looking ahead, the global drug delivery market is projected to grow substantially, with an estimated valuation of over $63 billion by 2032. This growth is driven by the rising demand for more effective delivery mechanisms, especially for biologics, which have traditionally struggled to reach intracellular targets due to their size and vulnerability to degradation.
🔸The Highest Recorded SPAM Fidelities of Any Quantum Computing Platform on the Market at 99.9993%
One of the biggest hurdles quantum computing faces today is error rates—especially when it comes to preparing, manipulating, and measuring qubits, which form the backbone of quantum operations. Low error rates are critical because even small mistakes can snowball and ruin the fragile quantum states needed for accurate computation. Errors can happen at any stage: state preparation, gate operations, or measurement. If left unchecked, these errors lead to incorrect results, making the quantum computer practically useless.
In a groundbreaking development for the field, Oxford Ionics, working alongside the Department of Physics at Oxford University, has set a new world record for quantum state preparation and measurement (SPAM) fidelity, achieving an unprecedented 99.9993%. This milestone marks the highest precision ever recorded in quantum computing and positions Oxford Ionics at the forefront of this rapidly advancing industry. They’ve cut SPAM errors by a factor of 13, thanks to a new protocol that detects and eliminates qubits in incorrect states, boosting readout accuracy.
So, why is this such a big deal? Oxford Ionics' advantage lies in their patented "Electronic Qubit Control" system, which, unlike traditional methods, uses electronics rather than lasers to execute quantum gate operations.
This sets Oxford Ionics apart from other ventures in quantum computing. Typically, lasers are used to manipulate quantum gates. While lasers work in research labs, they hit serious roadblocks when it comes to scaling up for commercial use. Oxford Ionics has sidestepped this challenge by replacing lasers with electronics to control qubits.
Here’s why that matters: In a world where scalability is everything, moving from laser-based systems to an all-electronic platform fundamentally changes the game. Electronics-based control can be integrated into existing semiconductor manufacturing processes. This means Oxford Ionics can replicate unit cells quickly and efficiently, using established supply chains. Scaling up quantum computers becomes less about overcoming new physics challenges and more about fine-tuning well-understood engineering practices.
For those interested in diving deeper into the technical details, check out their paper, High-fidelity Heralded Quantum State Preparation and Measurement.
This breakthrough isn't a one-off achievement. Oxford Ionics has also recently set world records for single- and two-qubit gate fidelities—two other critical metrics for evaluating quantum computing performance. Hitting record numbers across all three key benchmarks shows that Oxford Ionics is fast-tracking the development of a quantum platform that combines high performance with reliability. And that’s crucial if we want to unlock real-world applications.
This latest result is more than just a technical success. It’s a key step toward realizing the commercial and societal impacts of quantum computing. By drastically reducing error rates, the field is moving closer to solving problems that classical computers just can’t handle, like molecular simulations for drug discovery or optimizing complex systems like global supply chains.