Fact-checked by Grok 2 weeks ago

Deep tech

Deep tech refers to science-based technologies and innovations that emerge from breakthrough scientific research and engineering, often involving substantial uncertainty, long research and development cycles, and high capital requirements to create novel solutions for complex societal challenges.^[1] These technologies typically produce tangible products or processes, distinguishing them from digital or incremental innovations by their focus on physical advancements in areas like biotechnology, materials science, and energy systems.^[1] Key characteristics of deep tech include its positioning at the frontier of knowledge, strong ties to ecosystems such as universities and national laboratories, and a mission-driven orientation that aligns with 97% of ventures contributing to the United Nations Sustainable Development Goals.^[1] Deep tech ventures often originate from academic institutions, rely on elite talent from top-tier universities, and navigate a dynamic de-risking process involving iterative milestones to overcome the "valley of death" between invention and market viability.^[1] Unlike software-focused startups, they face elevated risks from technological feasibility, manufacturing scalability, regulatory approval, and market adoption, with development timelines extending 10 to 15 years and funding needs ranging from $20 million in early stages to over $1 billion for commercialization.^[2]^[3] Deep tech spans domains such as advanced materials and nanotechnology, biotechnology and agtech, novel artificial intelligence, quantum computing, robotics, clean energy, and space technologies, enabling scalable businesses that address pressing issues like climate change and healthcare.^[4] Its economic impact is profound, with one high-tech job generating up to five additional jobs and ecosystems fostering deep tech projected to create $1 trillion in enterprise value and 1 million jobs by 2030 in leading regions like Europe.^[4]^[3] Notable examples include BioNTech's mRNA platform for vaccines, Carbon Engineering's direct air capture for carbon removal, and Agility Robotics' bipedal robots for logistics, demonstrating how deep tech drives radical innovation and sustainability.^[2]

Fundamentals

Definition and Scope

Deep tech refers to advanced technologies that are fundamentally rooted in substantial scientific discoveries or engineering innovations, requiring intensive research and development (R&D) efforts and often taking several years to decades to mature into viable products or applications.^[5] Unlike incremental or "shallow" technologies, which focus on iterative improvements to existing systems—such as enhancements to user interfaces in software—deep tech involves paradigm-shifting breakthroughs that address complex, fundamental challenges with high technical risk and uncertainty.^[6] This distinction underscores deep tech's emphasis on non-trivial advancements rather than rapid, low-barrier deployments.^[1] The scope of deep tech is broadly defined by its reliance on cutting-edge science and engineering to enable transformative solutions, typically spanning multiple disciplines and demanding significant capital and expertise.^[7] It includes core fields such as artificial intelligence (AI), biotechnology and genomics, quantum computing, and advanced materials or nanotechnology, where innovations emerge from deep scientific understanding rather than off-the-shelf components.^[8] This excludes more conventional categories like consumer software applications or minor hardware iterations, which prioritize scalability and market fit over groundbreaking R&D.^[6] Representative examples illustrate these boundaries: quantum sensors, which leverage quantum entanglement for ultra-precise measurements in navigation or medical imaging, qualify as deep tech due to their foundation in quantum physics principles and prolonged development cycles.^[8] In contrast, standard mobile applications—such as social media tools built on existing APIs—do not, as they represent incremental digital enhancements without novel scientific underpinnings.^[7] The term "deep tech" was coined in 2014 by investor Swati Chaturvedi and gained prominence in the mid-2010s among venture capital circles to categorize high-barrier innovations that contrast with the era's dominant digital startups, emphasizing instead science-driven ventures with extended timelines and ecosystem dependencies.^[5]

Key Characteristics

Deep tech innovations are distinguished by their high technical complexity, which demands profound expertise in fields such as physics, biology, or mathematics, often stemming from academic or research laboratories where proprietary intellectual property (IP) is developed.^[1] These ventures operate at the scientific frontier, relying on unique technological advantages that are difficult to replicate, and involve intricate research and development (R&D) processes fraught with significant technical risks and uncertainties.^[1] Unlike consumer-facing software, deep tech requires interdisciplinary teams to navigate advanced engineering principles and experimental validation, contributing to its lower density in startup ecosystems compared to traditional tech.^[9] A hallmark of deep tech is its extended development timelines, typically spanning 7 to 15 years from initial lab discovery to commercial market entry, due to the iterative nature of scientific validation and regulatory approvals.^[1] This prolonged cycle is exacerbated by high failure rates; for instance, in biotechnology, over 90% of drug candidates fail during clinical development, driven by efficacy shortfalls, safety issues, and unforeseen biological complexities.^[10] Such durations contrast sharply with the rapid iteration seen in software, demanding sustained commitment from founders and investors to de-risk breakthroughs over multiple phases.^[1] Scalability in deep tech presents unique nonlinear challenges, constrained by physical, chemical, or biological limits that prevent the linear growth characteristic of digital technologies.^[11] Unlike software, which can scale via cloud infrastructure with marginal costs, deep tech often involves hardware or material-based solutions requiring complex manufacturing, supply chain orchestration, and regulatory certifications, making economic scaling a formidable engineering hurdle.^[1] These barriers can lead to exponential cost increases as production volumes rise, necessitating specialized facilities and expertise to overcome material or thermodynamic constraints.^[11] The capital intensity of deep tech is pronounced, with ventures frequently requiring upwards of $50 million in funding to reach prototype stages, and often hundreds of millions for full commercialization, sourced from venture capital firms, corporate investors, or government grants.^[12] This stems from the need for extensive R&D infrastructure, clinical trials, or fabrication facilities, far exceeding the seed-stage investments typical in software startups.^[1] Public funding plays a critical role, as private markets alone struggle to cover the multi-year burn rates without proven revenue.^[1] Deep tech exhibits a high-risk profile characterized by binary outcomes—either transformative breakthroughs or outright failure—owing to the probabilistic nature of scientific discovery and market adoption.^[1] Technological, commercial, and regulatory uncertainties amplify this, with success hinging on validation milestones that can pivot entire trajectories.^[1] Patents serve as core assets in mitigating these risks, providing defensible moats through exclusive rights to novel inventions, which are essential for attracting follow-on investment and deterring imitation in capital-scarce environments.^[1]

Historical Development

Origins in Scientific Breakthroughs

The origins of deep tech trace back to pivotal post-World War II scientific breakthroughs that established the foundational principles for advanced technologies in computing, materials science, and biotechnology. In 1947, researchers at Bell Laboratories developed the first point-contact transistor, a semiconductor device that amplified electrical signals and replaced bulky vacuum tubes, enabling the miniaturization of electronics essential to modern computing and information processing.^[13] This invention, demonstrated by John Bardeen and Walter Brattain using germanium, marked a shift toward solid-state physics that underpins deep tech innovations in hardware. Similarly, in 1953, James Watson and Francis Crick proposed the double-helix structure of DNA, revealing the molecular basis of genetic inheritance and laying the groundwork for genomics and synthetic biology.^[14] Their model, informed by X-ray diffraction data, demonstrated how DNA's complementary base pairing enables replication, a concept central to subsequent biotechnological advancements.^[15] Research institutions played a crucial role in fostering these discoveries through sustained investment in fundamental science. Bell Labs, established as part of the AT&T monopoly, provided interdisciplinary environments that produced the transistor and advanced quantum electronics in the late 1940s and 1950s.^[16] The Massachusetts Institute of Technology (MIT) contributed to computing foundations, developing magnetic core memory in the 1950s for reliable data storage in early digital systems and pioneering time-sharing concepts through Project MAC in the 1960s.^[17] ^[18] Meanwhile, Los Alamos National Laboratory, building on its Manhattan Project legacy, advanced nuclear physics and computational modeling during the 1950s and 1960s, including early supercomputing applications for simulating complex physical systems.^[19] From the 1950s to the 1970s, the focus on basic research transitioned toward practical applications, particularly through the application of quantum mechanics to materials like semiconductors. Theoretical advancements in band theory and electron behavior in solids, refined in the post-war era, enabled the engineering of semiconductor devices that manipulated quantum effects for amplification and switching. This period saw quantum formalization in solid-state contexts yield prototypes for integrated circuits, bridging abstract physics to technological prototypes without immediate commercial intent.^[13] Key pioneers embodied this era's innovative spirit. William Shockley, born in 1910 and a theoretical physicist at Bell Labs, co-invented the junction transistor in 1948, improving on the point-contact version and earning the 1956 Nobel Prize in Physics for enabling scalable semiconductor production.^[20] In genetics, James Watson, an American biologist born in 1928, collaborated with Francis Crick, a British physicist born in 1916, at the University of Cambridge; their 1953 insight into DNA's structure, recognized with the 1962 Nobel Prize, transformed biology from descriptive science to a predictive engineering discipline.^[21] These figures' work at elite institutions exemplified how targeted basic research seeded deep tech's core domains.

Evolution and Key Milestones

The evolution of deep tech from the 1980s onward shifted it from niche academic endeavors to a burgeoning industry sector, driven by breakthroughs in biotechnology and computing that required substantial scientific and engineering investment. In the late 1980s and 1990s, the biotech boom gained momentum with the launch of the Human Genome Project in 1990, an international collaboration coordinated by the U.S. Department of Energy and National Institutes of Health that aimed to map and sequence the human genome, culminating in a working draft by 2000 and full completion in 2003. This project not only advanced genomics but also spurred the development of sequencing technologies and bioinformatics tools essential for modern deep tech applications in healthcare. Paralleling these efforts, the initial discovery of clustered regularly interspaced short palindromic repeats (CRISPR) in bacterial genomes occurred in 1987 by Japanese researchers studying E. coli, marking early precursors to gene-editing systems that would later revolutionize biotechnology, with further characterizations emerging in the 1990s.^[22]^[23]^[24] The 2000s and 2010s saw deep tech mature through key advancements in artificial intelligence and quantum technologies, alongside the formal recognition of the sector by investors and accelerators. A landmark in AI came in 2012 when the AlexNet convolutional neural network, developed by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton at the University of Toronto, achieved a top-5 error rate of 15.3% in the ImageNet Large Scale Visual Recognition Challenge, outperforming previous methods by over 10 percentage points and igniting the deep learning revolution. This success demonstrated the scalability of neural networks on GPUs, paving the way for widespread commercial adoption in computer vision and beyond. By the mid-2010s, the term "deep tech" gained traction as programs like Y Combinator's 2014 Demo Day showcased cohorts focused on hardware, biotech, and science-driven startups solving grand challenges, signaling a pivot toward capital-intensive innovations over software alone. In quantum computing, Google's 2019 announcement of quantum supremacy—using its 53-qubit Sycamore processor to perform a random circuit sampling task in 200 seconds that would take a supercomputer 10,000 years—highlighted the field's progress toward practical utility, though debates persisted on the claim's scope. The 2019 European Deep Tech Report by Dealroom.co underscored the continent's rising prominence, documenting over €8 billion in deep tech investments in 2018 and identifying clusters in AI, biotech, and advanced manufacturing.^[25]^[26]^[27] The 2020s accelerated deep tech's integration into global economies, fueled by pandemic responses and surging investments, while expanding geographically. The COVID-19 crisis exemplified deep tech's real-world impact through mRNA vaccine platforms, with Moderna's mRNA-1273 and BioNTech/Pfizer's BNT162b2 receiving emergency authorization in late 2020 after leveraging synthetic biology and lipid nanoparticle delivery systems developed over decades. This success not only curbed the pandemic but validated mRNA as a versatile deep tech modality for future therapeutics. Global deep tech funding reached $79 billion in 2023, comprising about 24% of total venture capital and spanning AI, climate tech, and quantum domains, despite a broader VC downturn. Asia emerged as a key player, with government-led quantum initiatives such as China's prioritization of quantum information science in its 14th Five-Year Plan (2021–2025), allocating billions to R&D, and Singapore's launch of the National Quantum-Safe Network Plus in 2023 to secure communications against quantum threats. Into 2024 and 2025, deep tech continued its momentum, with the sector capturing approximately 33% of global venture capital funding in 2024 amid trends like agentic AI and post-quantum cryptography.^[28]^[29]^[30]^[31]^[32]^[33] These milestones underscore deep tech's transition to a trillion-dollar industry ecosystem.

Core Technologies

Artificial Intelligence and Machine Learning

Artificial intelligence (AI) and machine learning (ML) form a foundational pillar of deep tech, enabling systems to learn complex patterns from vast datasets and make predictions or decisions with human-like proficiency. At their core, neural networks mimic the human brain's structure, consisting of interconnected layers of nodes that process input data through weighted connections to produce outputs. These networks, particularly deep neural networks with multiple hidden layers, power advancements in deep tech by handling high-dimensional data that traditional algorithms cannot efficiently manage.^[34] A key enabler of modern neural networks is the backpropagation algorithm, introduced in 1986, which efficiently computes gradients to adjust network weights during training by propagating errors backward through the layers. This method revolutionized deep learning by making it feasible to train multi-layer networks on large datasets, overcoming earlier limitations in optimization. Building on this, the transformer architecture, proposed in 2017, relies on self-attention mechanisms to process sequential data in parallel, eliminating the need for recurrent structures and enabling scalable models for tasks like natural language processing. Transformers have become ubiquitous in deep tech, underpinning large-scale language models that drive innovation across computational domains.^[35]^[36] In deep tech contexts, reinforcement learning (RL) extends ML principles to enable autonomous decision-making in dynamic environments, such as robotics, where agents learn optimal actions through trial-and-error interactions with physical systems to maximize rewards. For instance, deep RL has achieved champion-level performance in drone racing by training policies that navigate complex obstacle courses at high speeds. Similarly, generative adversarial networks (GANs), introduced in 2014, pit a generator against a discriminator to produce realistic synthetic data, finding applications in drug discovery by generating novel molecular structures that accelerate hit identification in pharmaceutical research.^[37]^[38]^[39] Training these advanced models demands immense computational resources, heavily reliant on specialized hardware like graphics processing units (GPUs) from NVIDIA and tensor processing units (TPUs) from Google, which parallelize matrix operations essential for gradient descent. For example, models in the GPT series require approximately $10^{25} floating-point operations (FLOPs) for training, highlighting the scale of compute needed to achieve state-of-the-art performance. Model complexity is often measured by the number of parameters, with GPT-4 estimated at around 1.8 trillion, allowing it to capture intricate representations but necessitating efficient hardware to manage memory and inference.^[40]^[41]^[42] A landmark breakthrough illustrating AI's deep tech impact is AlphaFold, developed by DeepMind in 2020, which uses deep learning to predict protein structures with atomic accuracy, solving a decades-old biological challenge and accelerating research in genomics and drug design. This was further advanced with AlphaFold 3 in 2024, which predicts structures of protein complexes with DNA, RNA, ligands, and ions. By integrating evolutionary data with neural network architectures, AlphaFold demonstrates how ML can bridge computational and scientific frontiers, achieving median backbone accuracy improvements of over 50% on challenging targets compared to prior methods.^[43]^[44]

Biotechnology and Genomics

Biotechnology and genomics represent a cornerstone of deep tech, leveraging molecular biology and genetic engineering to manipulate biological systems at unprecedented precision. These fields enable the redesign of organisms and genomes, driving innovations from targeted therapies to sustainable biomanufacturing. Central to this domain are techniques that allow for programmable control over genetic material, transforming fundamental scientific discoveries into scalable technologies. A pivotal advancement in gene editing is the CRISPR-Cas9 system, introduced in 2012, which uses a guide RNA to direct the Cas9 enzyme for precise DNA cleavage and modification.^[45] This tool has revolutionized genomics by enabling efficient, cost-effective editing of genes in various organisms, facilitating research into disease mechanisms and therapeutic interventions. Complementing CRISPR, synthetic biology employs engineering principles to design novel organisms, exemplified by the 2010 creation of a synthetic bacterial genome that was transplanted into a recipient cell to produce a fully functional, self-replicating organism.^[46] These methods underscore deep tech's emphasis on modularity and standardization in biological design, allowing for the construction of custom genetic circuits and metabolic pathways. Genomics has advanced dramatically through next-generation sequencing (NGS), first commercialized around 2005, which parallelized DNA readout to drastically reduce costs.^[47] The price per human genome dropped from approximately $100 million in the early 2000s to about $200 as of 2025, democratizing access to large-scale genomic data and accelerating discoveries in personalized medicine and evolutionary biology.^[47]^[48] This cost reduction, driven by innovations in sequencing chemistry and instrumentation, has enabled routine whole-genome analysis, revealing insights into genetic variations and their functional impacts. Deep tech applications in biotechnology include chimeric antigen receptor (CAR) T-cell therapy, where patient T cells are genetically engineered to target cancer cells; the first such therapy, tisagenlecleucel (Kymriah), received FDA approval in 2017 for pediatric acute lymphoblastic leukemia.^[49] Another example is organoids—three-dimensional, stem cell-derived mini-organs used for drug testing—which mimic tissue architecture and physiology more accurately than traditional 2D cultures, as demonstrated in high-throughput screens for breast cancer therapeutics.^[50] These organoids allow for patient-specific modeling, improving prediction of drug efficacy and reducing reliance on animal models. Despite these breakthroughs, scaling biotechnological processes poses significant challenges, particularly in biomanufacturing where yields and purity rates can fall below 50% in some peptide synthesis and fermentation-based methods, requiring intensive optimization to achieve commercial viability.^[51] Recent CRISPR variants have addressed precision issues, achieving 80-90% reductions in off-target editing through modifications like zwitterionic coatings on Cas9, enhancing safety for therapeutic applications.^[52] These improvements highlight ongoing efforts to balance efficiency and specificity in deep tech genomics. Quantum computing represents a paradigm shift in information processing by leveraging principles of quantum mechanics, primarily through the use of qubits that can exist in a superposition of states, allowing them to represent multiple values simultaneously, unlike classical bits that are strictly 0 or 1. Superposition enables exponential computational scaling, as n qubits can encode 2^n states concurrently. Entanglement, another core phenomenon, links qubits such that the state of one instantly influences another, regardless of distance, facilitating correlated operations essential for quantum algorithms. A seminal demonstration of this power is Shor's algorithm, introduced in 1994, which efficiently factors large integers—a task intractable for classical computers—using quantum Fourier transforms to exploit periodicity in modular arithmetic.^[53] The current landscape of quantum computing is dominated by Noisy Intermediate-Scale Quantum (NISQ) devices, which operate with 50 to over 1,000 qubits but suffer from high error rates due to decoherence and imperfect gates.^[54] Typical two-qubit gate error rates hover around 0.1% to 1%, limiting practical utility to short-depth circuits without full error correction.^[55] Key milestones include IBM's Osprey processor, released in 2022 with 433 superconducting qubits, marking a tripling of scale from prior systems and enabling exploration of quantum advantage in specific simulations.^[56] Another landmark is China's Micius satellite, launched in 2016, which demonstrated quantum key distribution over 1,200 kilometers, validating space-based quantum communication protocols. Beyond computing, quantum technologies extend to sensing and cryptography, where entanglement and superposition enhance precision and security. Quantum sensors exploit coherent quantum states for measurements surpassing classical limits, such as in detecting minute gravitational perturbations; for instance, quantum squeezing techniques in interferometers like LIGO have pushed sensitivity beyond the standard quantum limit for gravitational wave detection.^[57] In quantum cryptography, the BB84 protocol, proposed in 1984, uses polarized photons to distribute secure keys, detecting eavesdroppers via the no-cloning theorem since any interception disturbs the quantum states. Achieving fault-tolerant quantum computing, capable of running complex algorithms reliably, demands scaling to over one million physical qubits to implement error-correcting codes that suppress errors below the fault-tolerance threshold.^[58] Current NISQ systems serve as testbeds for hybrid quantum-classical algorithms, with brief intersections to fields like artificial intelligence for tasks such as variational quantum eigensolvers. The path forward involves iterative improvements in qubit fidelity and connectivity to realize universal quantum computation.

Advanced Materials and Nanotechnology

Advanced materials and nanotechnology represent a cornerstone of deep tech, enabling the engineering of substances at the atomic and molecular scales to achieve properties unattainable with conventional materials. These innovations focus on manipulating matter to create structures with exceptional mechanical, electrical, and optical characteristics, underpinning hardware advancements in computing, sensing, and energy systems. Key concepts include graphene, a single layer of carbon atoms isolated in 2004 by Andre Geim and Konstantin Novoselov using mechanical exfoliation from graphite, which exhibits remarkable electron mobility and thermal conductivity.^[59] This breakthrough earned them the 2010 Nobel Prize in Physics for groundbreaking experiments on the two-dimensional material.^[60] Metamaterials, artificially structured composites, further exemplify these advances by exhibiting properties such as negative refractive indices, enabling applications like electromagnetic cloaking and superlenses. Cloaking devices, which bend light around objects to render them invisible, rely on metamaterials designed to guide waves without scattering, as demonstrated in early theoretical and experimental works.^[61] Superlenses, overcoming the diffraction limit of traditional optics, use negative-index metamaterials to achieve subwavelength resolution, with prototypes resolving features as small as 60 nm at microwave frequencies.^[62] In nanotechnology, carbon nanotubes (CNTs) highlight the potential for superior mechanical performance, with single-walled variants possessing tensile strengths up to 100 GPa—approximately 100 times that of steel—due to their seamless cylindrical structure of sp²-bonded carbon atoms.^[63] Self-assembling nanostructures, where molecules spontaneously organize into ordered architectures through non-covalent interactions like hydrogen bonding and van der Waals forces, offer scalable routes to complex patterns, such as nanoparticle ensembles forming photonic crystals with tunable optical properties.^[64] Deep tech applications leverage these materials in hardware, particularly two-dimensional (2D) semiconductors like molybdenum disulfide (MoS₂) for next-generation transistors. Monolayer MoS₂ transistors have demonstrated high on/off ratios exceeding 10⁶ and mobilities around 200 cm²/V·s, enabling ultra-scaled devices beyond silicon limits for flexible electronics and high-frequency circuits.^[65] These 2D materials address short-channel effects in transistors, supporting gate lengths below 5 nm while maintaining low power dissipation. Synthesis methods, such as chemical vapor deposition (CVD), are critical for producing large-scale nanomaterials with controlled quality. In CVD, hydrocarbon precursors like methane decompose on metal catalysts (e.g., copper) at 1000°C, yielding uniform graphene films up to wafer-sized with monolayer coverage over 95% of the area.^[66] High-quality graphene from optimized CVD achieves purity levels above 99%, minimizing defects like grain boundaries that degrade electrical performance, though yields depend on precursor purity and growth parameters.^[67] Milestones in this field include the 2010 Nobel recognition for graphene and the emergence of nanobot prototypes in the 2020s, showcasing integrated nanoscale machines for precise manipulation. These prototypes, often propelled by chemical or magnetic mechanisms, demonstrate autonomous assembly and movement at scales below 100 nm, paving the way for hardware in sensing and assembly tasks.^[68]

Applications and Societal Impact

Healthcare and Life Sciences

Deep tech is revolutionizing healthcare and life sciences by integrating advanced artificial intelligence, biotechnology, and nanotechnology to enhance diagnostics, develop novel therapeutics, and enable personalized medicine, ultimately improving patient outcomes and accelerating medical innovation. In diagnostics, AI-driven deep learning models analyze medical imaging with high precision, enabling earlier and more accurate detection of diseases such as cancer. For instance, convolutional neural networks applied to histopathological images have achieved approximately 95% accuracy in classifying metastatic breast cancer cells versus benign ones, outperforming traditional methods in speed and reliability.^[69] Therapeutics in this domain leverage gene editing and nanoscale engineering for targeted interventions. Gene therapies like Zolgensma (onasemnogene abeparvovec), approved by the FDA in 2019 for spinal muscular atrophy, represent a breakthrough by delivering a functional SMN1 gene via adeno-associated virus vectors, offering a one-time treatment that halts disease progression in infants. Priced at $2.125 million per dose, it underscores the high-impact potential of deep tech despite cost challenges. Complementing this, nanobots facilitate precise drug delivery to tumor sites; for example, DNA nanorobots designed to release thrombin specifically at tumor-associated blood vessels induce localized thrombosis, starving cancer cells of oxygen and minimizing off-target effects in preclinical models.^[70]^[71]^[72] Personalized medicine tailors treatments using genomic data, with pharmacogenomics guiding drug selection to mitigate risks. Implementation of multi-gene pharmacogenetic panels has reduced clinically relevant adverse drug reactions by 30% in hospital settings, allowing adjustments based on genetic variants affecting drug metabolism, such as CYP2C19 for clopidogrel. This approach enhances efficacy while lowering hospitalization rates from drug toxicities. Broader impacts include shortened clinical timelines through AI-powered simulations; AI-powered simulations can compress development timelines by up to 12 months per asset through optimized trial design and patient stratification.^[73] A landmark case study is the rapid development of mRNA vaccines by Pfizer and BioNTech in 2020, which demonstrated deep tech's scalability during the COVID-19 pandemic. The BNT162b2 vaccine, utilizing lipid nanoparticle-encapsulated mRNA to encode the SARS-CoV-2 spike protein, achieved 95% efficacy in phase 3 trials and received emergency use authorization within months, saving millions of lives and paving the way for future mRNA-based therapies against infectious diseases and cancers. This success highlights how foundational biotech advancements, briefly intersecting with AI for sequence optimization, can address global health crises efficiently.

Energy and Sustainability

Deep tech innovations are transforming the energy sector by enabling more efficient renewable energy generation. Perovskite solar cells, leveraging advanced materials science, have achieved power conversion efficiencies exceeding 25% by 2023, surpassing traditional silicon-based cells and paving the way for cost-effective, scalable photovoltaics.^[74] In nuclear fusion, prototypes like the International Thermonuclear Experimental Reactor (ITER) marked significant progress in 2025 by completing assembly of the world's most powerful magnet, a critical step toward demonstrating sustained fusion reactions for clean, limitless energy.^[75] Energy storage advancements driven by deep tech address intermittency in renewables through next-generation batteries and alternative methods. Solid-state batteries have reached energy densities of up to 600 Wh/kg in prototypes, more than doubling the 250 Wh/kg typical of conventional lithium-ion batteries, enabling longer-range electric vehicles and grid-scale storage with enhanced safety.^[76] Nanomaterials, such as titanium dioxide nanoparticles, facilitate efficient hydrogen electrolysis by catalyzing water splitting with reduced overpotentials, such as 470 mV to reach 10 mA/cm² in neutral conditions and supporting green hydrogen as a versatile energy carrier.^[77] In climate technology, artificial intelligence optimizes material design for carbon capture, identifying metal-organic frameworks and membranes that achieve over 90% CO2 absorption from flue gases while minimizing energy penalties.^[78] These AI-driven approaches accelerate discovery, reducing development timelines from years to months. Deep tech's broader integration into energy systems holds substantial potential to curb global emissions; models indicate that advanced climate technologies could contribute to 20-30% reductions by 2050, aligning with IPCC pathways for net-zero targets through enhanced efficiency and deployment.^[79] A notable case study involves quantum computing for grid management, where hybrid quantum-classical algorithms optimize power flow in simulated networks like the IEEE 14-bus system, reducing energy losses by up to 26% compared to classical methods.^[80] This demonstrates quantum tech's capacity to minimize transmission inefficiencies, supporting resilient, low-loss electricity distribution amid rising renewable integration.

Transportation and Manufacturing

Deep tech innovations are transforming transportation by enabling autonomous systems that operate at advanced levels of self-driving capability. Companies like Waymo have deployed fully driverless robotaxis at SAE Level 4 autonomy, providing public rides without human intervention in geofenced areas such as Phoenix, Arizona, since December 2020, with expansions to cities like San Francisco and Los Angeles throughout the 2020s.^[81] These systems leverage artificial intelligence for real-time perception, decision-making, and navigation, reducing human error and enabling scalable urban mobility solutions. While full SAE Level 5 autonomy—requiring no operational restrictions—remains aspirational, ongoing developments in sensor fusion and machine learning are bridging the gap toward unrestricted operations.^[82] In manufacturing, advanced 3D printing techniques incorporating nanomaterials are revolutionizing aerospace component production by allowing complex geometries that traditional methods cannot achieve. For instance, additive manufacturing with nanocomposites enables the creation of lightweight lattice structures for aircraft parts, achieving up to 50% weight reduction compared to conventional aluminum components while maintaining structural integrity.^[83] This not only lowers fuel consumption but also reduces material waste during fabrication, as demonstrated in Boeing's use of 3D-printed titanium parts for the 787 Dreamliner, which cut production time and assembly complexity.^[84] Such applications draw on nanotechnology for enhanced material properties, like improved strength-to-weight ratios in carbon nanotube-infused polymers.^[85] Electric and hybrid propulsion systems benefit from deep tech advancements in battery technology, particularly solid-state batteries, which promise higher energy density and safety over liquid lithium-ion alternatives. Developments by Toyota and Samsung SDI target ranges exceeding 500 miles per charge, with prototypes demonstrating 600-mile capabilities and rapid 9-minute charging times.^[86] These batteries replace flammable electrolytes with solid materials, mitigating risks like thermal runaway and enabling lighter, more compact designs for electric vehicles (EVs). In logistics, drone swarms coordinated via AI algorithms facilitate efficient package delivery, with examples showing up to 30% cost reductions through parallel operations in urban and rural settings.^[87] Efficiency gains in manufacturing and supply chains are amplified by robotics and machine learning integration. Robotic automation in assembly lines has reduced defect rates by approximately 40%, as seen in BMW's implementation of AI-powered vision systems for quality inspection, minimizing flaws in high-precision parts.^[88] Similarly, machine learning optimizes supply chain operations by forecasting demand and routing, yielding up to 40% reductions in overall costs through predictive analytics that prevent stockouts and overstock.^[89] These tools enable just-in-time production, cutting waste and enhancing responsiveness to market fluctuations. A notable case study in deep tech transportation is the Hyperloop concept, first proposed by Elon Musk in 2013 as a high-speed system using near-vacuum tubes to minimize air resistance.^[90] The design integrates magnetic levitation for pod propulsion and low-pressure environments to achieve speeds over 700 mph, with control systems managing alignment and stability. Subsequent developments by organizations like Hyperloop Transportation Technologies have incorporated AI for autonomous pod navigation and real-time safety monitoring, addressing challenges in vacuum maintenance and traffic coordination.^[91]

Challenges and Future Outlook

Technical and Ethical Hurdles

Deep tech innovations, while transformative, face significant technical hurdles that impede their practical deployment. Scalability remains a primary challenge, particularly in quantum computing, where quantum decoherence—the loss of quantum coherence due to environmental interactions—severely limits computation times. Current qubit coherence times typically range from microseconds to milliseconds, with advanced superconducting qubits reaching up to about 1.7 milliseconds as reported in 2025, requiring algorithms to complete before decoherence disrupts the quantum state.^[92]^[93] Integration across deep tech fields, such as combining artificial intelligence with biotechnology or quantum systems, introduces further complexities, including incompatible hardware architectures, data interoperability issues, and the need for specialized interdisciplinary expertise to harmonize disparate technologies.^[94] Ethical concerns compound these technical barriers, especially in artificial intelligence and biotechnology. In AI decision-making systems, biases inherent in training data can amplify errors for underrepresented groups; for instance, algorithmic bias in healthcare AI has been shown to result in 17% lower diagnostic accuracy for minority patients compared to majority groups.^[95] Biotechnology advancements like CRISPR/Cas9 gene editing carry dual-use risks, enabling both therapeutic applications and potential misuse for creating bioweapons, such as engineered pathogens that could cause widespread harm.^[96] Privacy and equity issues exacerbate these dilemmas in genomics and related fields. Large-scale genomic datasets are often controlled by a few corporations and institutions, creating data monopolies that restrict access and stifle collaborative innovation while raising concerns over individual privacy breaches.^[97] Access gaps to advanced therapies, such as gene therapies, disproportionately affect low- and middle-income countries (LMICs), which bear nearly 90% of the global disease burden but host only a fraction of approved treatments—out of approximately 36 gene therapies approved worldwide as of mid-2025, only a small fraction (around 5) were available in LMICs, benefiting primarily high-income populations comprising about 16% of the world.^[98]^[99]^[100] Regulatory frameworks lag behind these rapid developments, creating gaps in oversight. For quantum encryption, there is a notable absence of standardized protocols to counter the "harvest now, decrypt later" threat, where adversaries collect encrypted data today for future quantum decryption, prompting calls for global migration to post-quantum cryptography by 2030.^[101]^[102] The European Union's AI Act, enacted in 2024, represents a key response by classifying AI systems by risk levels and mandating transparency and accountability measures to address ethical lapses.^[103] To mitigate these hurdles, strategies include establishing interdisciplinary ethics boards to oversee development and deployment, as exemplified by IBM's AI Ethics Board, which integrates diverse expertise to align technologies with societal values.^[104] Open-source initiatives also play a crucial role, fostering transparency and broader access; for example, projects like Data Hazards provide vocabularies for identifying ethical risks in data-intensive deep tech, enabling collaborative hazard mitigation across communities.^[105]

Funding, Commercialization, and Policy

The funding landscape for deep tech has seen substantial growth, with venture capital investments reaching approximately $80 billion globally in 2023, down from peaks of $160 billion in 2021 but still representing a significant share of overall tech funding amid economic pressures. In 2024, global tech VC rebounded to over $337 billion, with deep tech maintaining around 20-25% share in high-potential areas.^[106]^[107] This investment surge is driven by high-potential areas like artificial intelligence, biotechnology, and quantum computing, where deep tech captured about 20% of total VC activity by 2023, up from 10% a decade earlier.^[106] Government support has complemented private funding, exemplified by the U.S. CHIPS and Science Act of 2022, which allocates $52.7 billion to bolster domestic semiconductor manufacturing, research, and workforce development, addressing supply chain vulnerabilities in deep tech hardware, with implementations ongoing into 2025.^[108] Similarly, the European Union's Horizon Europe program provides €95.5 billion (approximately $100 billion) from 2021 to 2027 for research and innovation, prioritizing deep tech initiatives in areas like advanced materials and sustainable energy.^[109] Commercialization pathways in deep tech often begin with university spin-offs, which form a cornerstone of innovation transfer; for instance, university-affiliated ventures account for around 60% of the top-funded pharmaceutical and biotech startups by investment value.^[110] These spin-offs leverage academic research to prototype and scale technologies, with intellectual property (IP) valuation playing a critical role through established models such as the income approach (projecting future revenue streams from IP), the market approach (comparing to similar asset sales), and the cost approach (accounting for R&D expenditures).^[111] Effective IP strategies enable licensing deals or equity stakes, facilitating transitions from lab to market, though they require balancing exclusivity with collaborative ecosystems to attract partners. Policy frameworks significantly shape deep tech development, with national strategies fostering competitiveness while navigating geopolitical risks. The EU's Horizon Europe emphasizes collaborative R&D across borders to accelerate commercialization, funding over 15,000 projects by early 2025 in deep tech domains like quantum and biotech.^[112] In the U.S., the CHIPS Act not only injects capital but also imposes restrictions on technology transfers to adversaries, reflecting broader trade tensions in quantum computing where U.S.-China rivalry has led to export controls on sensitive hardware and algorithms to safeguard national security.^[113] These policies aim to mitigate risks like technology leakage, with international agreements emerging to standardize ethical guidelines for quantum advancements. A major barrier to deep tech commercialization is the "Valley of Death," the funding gap between early-stage research and market-ready scaling, where traditional venture capital often withdraws due to high risks and long timelines; estimates suggest only about 10% of deep tech ventures successfully bridge this phase.^[114] This challenge contributes to a low overall survival rate, with around 90% of startups failing to reach profitability, exacerbated by the need for substantial capital in capital-intensive fields like advanced materials.^[115] Despite these hurdles, successes like Moderna's 2018 initial public offering—raising a record $604 million as the largest biotech IPO at the time—demonstrate viable paths, propelled by strategic partnerships and IP protection that enabled rapid scaling during the COVID-19 pandemic.^[116] Emerging trends in deep tech funding include the rise of corporate venture arms, such as GV (formerly Google Ventures), which has invested in over 500 deep tech companies since 2009, focusing on AI, life sciences, and frontier technologies to integrate innovations into parent ecosystems.^[117] Additionally, impact investing is gaining traction for sustainability-focused deep tech, with funds targeting climate solutions that generated over $32 billion in 2023, emphasizing measurable environmental outcomes alongside financial returns through specialized vehicles like technology impact funds.^[118] These trends signal a maturing ecosystem, blending strategic corporate interests with global priorities like net-zero transitions.

References

[1]
[PDF] What is "Deep Tech" and what are Deep Tech Ventures? - MIT REAP
Two main approaches have emerged in the literature to define the concept of deep tech: the material approach, based on the sector to which the technology ...
[2]
[PDF] What is “Deep Tech” and why should corporate innovators care?
Jun 14, 2024 · Deep tech ventures are first-of-a-kind, meaning that they produce technical inventions based on cutting-edge research. As such, they are often ...
[3]
What are deep-tech startups?
Oct 24, 2025 · All that said, when we talk about deep-tech firms we mean those that are developing innovative solutions based on new technologies and science.What Are ``deep Tech'' Firms... · Why Do Startups, Early-Stage... · So, Why Is This All...
[4]
Europe's deep-tech engine could spur $1 trillion in economic growth
Oct 29, 2025 · Deep tech is a dynamic and ever-evolving sector in which some technologies that were once considered cutting edge are now becoming mainstream.
[5]
What is deep tech? | Definition from TechTarget
Dec 7, 2023 · Deep technology, or deep tech, refers to advanced technologies based on some form of substantial scientific or engineering innovation.
[6]
Deep Tech and Innovations | United Nations Development Programme
Deep tech is cutting-edge, disruptive technology built on scientific discoveries, with potential to radically transform industries, economies, and lives.
[7]
Deep Tech: The New Wave - IDB Publications
DeepTech startups are companies based on a scientific discovery or meaningful engineering innovation. Deep Tech startups involve significant technological risk ...
[8]
What Is Deep Tech? | Built In
Deep tech is a term used to describe highly sophisticated technology rooted in advanced scientific principles and engineering innovations.
[9]
European deep tech: What investors and corporations need to know
Dec 17, 2024 · While the term “deep tech” refers to technologies grounded in significant scientific and engineering challenges, it is essential to ...
[10]
Why 90% of clinical drug development fails and how to improve it?
Despite this validated effort, the overall success rate of clinical drug development remains low at 10%–15%5, 6, 7. Such persistent high failure rate raises ...Missing: credible | Show results with:credible
[11]
State of Deep Tech - Bessemer Venture Partners
Jun 14, 2023 · Common deep tech challenges · 1. Technology. Deep tech companies must overcome the challenge of making their technology work and scaling it ...Introducing The Xb100 · Deep Tech Benchmarks And... · Common Deep Tech ChallengesMissing: characteristics | Show results with:characteristics
[12]
[PDF] An Investor's Guide to Deep Tech - Boston Consulting Group
Venture capital funding of deep tech fell back from a re- cord $160 billion in 2021 to about $105 billion in 2022 and to $40 billion for the first half of 2023 ...
[13]
1947: Invention of the Point-Contact Transistor | The Silicon Engine
In December 1947, John Bardeen and Walter Brattain achieved transistor action using germanium with two gold contacts, amplifying a signal up to 100 times.
[14]
1953: DNA Double Helix
Apr 23, 2013 · Francis Crick and James Watson described the double helix structure of DNA. By the time Watson and Crick turned their attention to solving the chemical ...
[15]
The Discovery of the Double Helix, 1951-1953 | Francis Crick
The discovery in 1953 of the double helix, the twisted-ladder structure of deoxyribonucleic acid (DNA), by James Watson and Francis Crick marked a milestone ...
[16]
History | Nokia.com
In 1947, Bell Labs researchers invented the transistor to replace vacuum tubes in the telephone network. That invention, however, would transform the world, ...
[17]
Research at MIT - MIT Facts
1950s. Building the magnetic core memory that made digital computers possible. 1960s. Developing the inertial guidance systems for the Apollo space program.Missing: materials | Show results with:materials
[18]
Project MAC - Multics
Aug 14, 2014 · The CSAIL timeline documents the history of computing research at MIT, from the 1940s to the present. MIT CSAIL also provides a useful web page ...<|separator|>
[19]
Innovation Through the Years | Los Alamos National Laboratory
LANL designed the atomic bombs used to bring history's most violent and devastating conflict, World War II, to an abrupt end. Since the war, LANL has ensured ...Missing: 1950s- | Show results with:1950s-
[20]
William B. Shockley – Biographical - NobelPrize.org
William Shockley was born in London, England, on 13th February, 1910, the son of William Hillman Shockley, a mining engineer born in Massachusetts and his wife ...
[21]
Francis Crick, Rosalind Franklin, James Watson, and Maurice Wilkins
Sep 8, 2025 · In 1953, Watson and Crick published a short, strongly worded article in Nature announcing the double helix. Nine years later, Watson, Crick, and ...
[22]
Human Genome Project Timeline
Jul 5, 2022 · Completed in April 2003, the Human Genome Project gave us the ability to read nature's complete genetic blueprint for a human.
[23]
Human Genome Project Timeline
The Human Genome Project (HGP) refers to the international 13-year effort, formally begun in October 1990 and completed in 2003, to discover all the estimated ...
[24]
CRISPR–Cas9: A History of Its Discovery and Ethical ... - NIH
Aug 15, 2022 · 1987 – Short DNA repeats, later called CRISPR, were first noticed in bacterial genomes, and, in 1995, also found in archaea.
[25]
New At Y Combinator: Startups Solving Huge Problems! - TechCrunch
Aug 19, 2014 · During today's Y Combinator Demo Day, there was a notable block among the startups presenting that didn't give off the vibe we're used to ...
[26]
Quantum supremacy using a programmable superconducting ...
Oct 23, 2019 · The promise of quantum computers is that certain computational tasks might be executed exponentially faster on a quantum processor than on a ...
[27]
100+ European Deep Tech startups that raised funding in 2019
Feb 27, 2019 · Deep Tech investment continues to gain steam in Europe. In 2018, over €8 billion was invested in European startups with a Deep Tech edge.
[28]
The tangled history of mRNA vaccines - Nature
Sep 14, 2021 · Hundreds of scientists had worked on mRNA vaccines for decades before the coronavirus pandemic brought a breakthrough.Missing: deep | Show results with:deep
[29]
[PDF] State and trends of deep tech investment in 2023
Includes all companies worldwide that have completed Series B financing and subsequent rounds, accounting for more than 90 % of total funding.
[30]
A Deep Dive Into China's Five-Year, Deep Tech Vision
Nov 1, 2025 · China's 14th Five-Year Plan elevates quantum information as a core part of the nation's broader deep tech ecosystem.Missing: 2020s | Show results with:2020s
[31]
https://www.channelnewsasia.com/singapore/quantum-physics-computer-technology-sector-helios-quantinuum-5453116
[32]
[PDF] “FOR FOUNDATIONAL DISCOVERIES AND INVENTIONS THAT ...
Oct 8, 2024 · Inspired by biological neurons in the brain, ANNs are large collections of “neurons”, or nodes, connected by “synapses”, or weighted ...<|control11|><|separator|>
[33]
Learning representations by back-propagating errors - Nature
Oct 9, 1986 · Cite this article. Rumelhart, D., Hinton, G. & Williams, R. Learning representations by back-propagating errors. Nature 323, 533–536 (1986).
[34]
[1706.03762] Attention Is All You Need - arXiv
Jun 12, 2017 · We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely.
[35]
Champion-level drone racing using deep reinforcement learning
Aug 30, 2023 · Here we introduce Swift, an autonomous system that can race physical vehicles at the level of the human world champions.
[36]
[1406.2661] Generative Adversarial Networks - arXiv
Jun 10, 2014 · We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models.
[37]
Integrating artificial intelligence into small molecule development for ...
Oct 1, 2025 · GANs, on the other hand, employ a competitive learning framework between two networks: a generator that creates candidate molecules and a ...
[38]
Tensor Processing Units (TPUs) - Google Cloud
Google Cloud TPUs are custom-designed AI accelerators, which are optimized for training and inference of AI models. They are ideal for a variety of use cases, ...
[39]
Over 30 AI models have been trained at the scale of GPT-4
Jan 30, 2025 · The largest AI models today are trained with over 1025 floating-point operations (FLOP) of compute. The first model trained at this scale was ...
[40]
GPT-4 Architecture, Infrastructure, Training Dataset, Costs, Vision ...
Jul 10, 2023 · For example at batch size of 256 or 512, there are 512 FLOP/s or 1024 FLOP/s for each byte of memory that read in. This ratio more closely ...
[41]
Highly accurate protein structure prediction with AlphaFold - Nature
Jul 15, 2021 · Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure ...
[42]
A Programmable Dual-RNA–Guided DNA Endonuclease ... - Science
Jun 28, 2012 · Our study reveals a family of endonucleases that use dual-RNAs for site-specific DNA cleavage and highlights the potential to exploit the system for RNA- ...
[43]
Creation of a Bacterial Cell Controlled by a Chemically Synthesized ...
May 20, 2010 · We report the design, synthesis, and assembly of the 1.08–mega–base pair Mycoplasma mycoides JCVI-syn1.0 genome starting from digitized genome sequence ...
[44]
DNA Sequencing Costs: Data
May 16, 2023 · Data used to estimate the cost of sequencing the human genome over time since the Human Genome Project.
[45]
FDA approves tisagenlecleucel for B-cell ALL and tocilizumab for ...
Sep 7, 2017 · On August 30, 2017, the U.S. Food and Drug Administration granted regular approval to tisagenlecleucel (KYMRIAH, Novartis Pharmaceuticals Corp.)
[46]
Rapid drug screen using 3D tumor organoids - Science
Jan 10, 2018 · This study demonstrates the potential of breast cancer organoids as a functional high-throughput drug screen platform for personalized medicine.Missing: example | Show results with:example
[47]
Peptide Purity & Yield Optimizing in SPPS | PurePep Blog
Nov 4, 2020 · We look at ways to ensure optimal purity and yield in solid-phase peptide synthesis (SPPS).Missing: biomanufacturing less
[48]
Minimizing the off-target frequency of the CRISPR/Cas9 system via ...
Jun 14, 2023 · Results show that the off-target efficiency of zwitterlated CRISPR/Cas9 is reduced on average by 70% and can be as high as 90% when compared ...
[49]
Algorithms for quantum computation: discrete logarithms and factoring
This paper gives Las Vegas algorithms for finding discrete logarithms and factoring integers on a quantum computer that take a number of steps which is ...
[50]
[1801.00862] Quantum Computing in the NISQ era and beyond - arXiv
Jan 2, 2018 · NISQ devices will be useful ... Access Paper: View a PDF of the paper titled Quantum Computing in the NISQ era and beyond, by John Preskill.
[51]
Evidence for the utility of quantum computing before fault tolerance
Jun 14, 2023 · It is almost universally accepted that advanced quantum algorithms such as factoring or phase estimation will require quantum error correction.
[52]
IBM Unveils 400 Qubit-Plus Quantum Processor and Next ...
Nov 9, 2022 · The new 433 qubit 'Osprey' processor brings us a step closer to the point where quantum computers will be used to tackle previously unsolvable problems.
[53]
LIGO Surpasses the Quantum Limit | LIGO Lab | Caltech
Oct 23, 2023 · LIGO researchers report a significant advance in a quantum technology called squeezing that allows them to skirt around this limit and measure undulations in ...
[54]
[2406.06015] Superconducting qubits at the utility scale - arXiv
Jun 10, 2024 · This paper presents an architectural model of a potential FTQC based on superconducting qubits, divided into discrete modules and interconnected via coherent ...
[55]
Electric Field Effect in Atomically Thin Carbon Films - Science
Oct 22, 2004 · Graphene is the name given to a single layer of carbon atoms densely packed into a benzene-ring structure, and is widely used to describe ...Missing: isolation | Show results with:isolation
[56]
The Nobel Prize in Physics 2010 - NobelPrize.org
The Nobel Prize in Physics 2010 was awarded jointly to Andre Geim and Konstantin Novoselov for groundbreaking experiments regarding the two-dimensional ...
[57]
Electromagnetic cloaking with metamaterials - ScienceDirect.com
In this paper we review the basic principles of metamaterials, especially those for cloaking applications, and describe the recent developments in the field of ...
[58]
Subwavelength resolution with a negative-index metamaterial ...
Jun 19, 2007 · The authors report an impedance-matched, low loss negative-index metamaterial superlens that is capable of resolving subwavelength features of a ...Missing: original | Show results with:original
[59]
Strength of carbon nanotubes depends on their chemical structures
Jul 10, 2019 · Single-walled carbon nanotubes theoretically possess ultimate intrinsic tensile strengths in the 100–200 GPa range, among the highest in existing materials.Missing: steel | Show results with:steel
[60]
Properties and emerging applications of self-assembled structures ...
Dec 24, 2009 · Self-assembly has emerged as a powerful technique for controlling the structure and properties of ensembles of inorganic nanoparticles.
[61]
Breakdown of High-Performance Monolayer MoS2 Transistors
Oct 5, 2012 · MoS2 can effectively amplify signals and that it could be used for advanced analog circuits based on 2D materials. (c) 2012 American Institute ...
[62]
Chemical Vapour Deposition of Graphene—Synthesis ...
This review details the leading CVD methods for graphene growth, including hot-wall, cold-wall and plasma-enhanced CVD.
[63]
Effects of precursors' purity on graphene quality - AIP Publishing
Apr 9, 2020 · In this work, we show that the graphene defect density can be engineered during CVD synthesis using CH4 precursors containing different amounts ...
[64]
Propulsion mechanisms of micro/nanorobots: a review
This paper reviews the research progress of MNR in terms of propulsion mechanisms, and points out the pros and cons of different stimulus types.
[65]
CNN‐based deep learning approach for classification of invasive ...
Aug 30, 2024 · The results demonstrate an excellent performance ensuring an average of 95% accuracy in classifying metastatic cells against benign ones and 89% ...
[66]
[PDF] May 24, 2019 Summary Basis for Regulatory Action - ZOLGENSMA
May 24, 2019 · An ongoing Phase 3 clinical trial and a completed Phase 1 clinical trial provide the primary evidence of safety and effectiveness for the ...Missing: cost | Show results with:cost
[67]
Onasemnogene Abeparvovec for Spinal Muscular Atrophy - NIH
The treatment is going to cost $2.125 million (approximately 14 crore rupees) and the drug is given as one-time infusion over 1 h.[1] This makes the drug to be ...
[68]
A DNA nanorobot functions as a cancer therapeutic in ... - PubMed
We demonstrate that intravenously injected DNA nanorobots deliver thrombin specifically to tumor-associated blood vessels and induce intravascular thrombosis.<|control11|><|separator|>
[69]
Performance in clinical development with AI & ML | McKinsey
Jan 9, 2025 · AI-driven site selection can accelerate enrollment and reduce total trial duration across therapeutic areas. Enhanced trial management with an ...
[70]
Recent progress in the development of high-efficiency inverted ...
May 5, 2023 · The power conversion efficiency (PCE) of PSCs has increased to an impressive value of 25.7% after more than ten years of development. PSCs are a ...
[71]
Global nuclear fusion project crosses milestone with world's most ...
May 1, 2025 · A much-delayed nuclear fusion project involving more than 30 countries is ready to assemble the world's most powerful magnet - a key part of ...
[72]
German team creates new solid-state EV battery with 600 Wh/kg ...
Oct 30, 2025 · German researchers have developed a new solid-state lithium-sulfur battery that reduces electrolyte content and boosts energy density.
[73]
Efficient hydrogen evolution via neutral water electrolysis using ...
May 8, 2025 · Various transition metal oxides have been explored as promising candidates for hydrogen evolution reactions (HER), attracting considerable ...
[74]
Artificial intelligence and material design in carbon capture and ...
Ultimately, efficient optimization strategies should improve the conversion rate of CO2 into valuable chemicals, such as methanol and carbon monoxide (Tawalbeh ...
[75]
Digital Tech Can Reduce Emissions by up to 20% in High-Emitting ...
May 24, 2022 · Digital technologies can reduce greenhouse gas emissions by up to 20% by 2050 in the three highest-emitting sectors: energy, mobility and materials.Missing: deep 20-30%
[76]
[PDF] Hybrid Quantum-Classical Optimization Algorithms for Energy ...
The experimental results demonstrate that the hybrid approach outperforms traditional methods, with up to 26% energy loss reduction in the IEEE 14-bus system ...
[77]
Self-Driving Cars Gain Momentum in US - S&P Global
Sep 9, 2024 · US autonomous vehicle industry accelerates with Waymo leading and Cruise rebuilding amid technological breakthroughs.Self-Driving Cars Gain... · Waymo: Still The... · Cruise AutomationMissing: SAE 2020s
[78]
The 6 Levels of Vehicle Autonomy Explained | Synopsys Automotive
Feb 15, 2025 · Level 5 cars won't even have steering wheels or acceleration/braking pedals. They will be free from geofencing, able to go anywhere and do ...Missing: 2020s | Show results with:2020s
[79]
Aerospace 3D Printing: Up to 50% Aircraft Weight Reduction - 3DX
Discover how aerospace 3D printing revolutionizes aircraft design, enabling up to 50% weight reduction. Explore benefits like enhanced fuel efficiency ...
[80]
Metal additive manufacturing in aerospace: A review - ScienceDirect
Nov 1, 2021 · This paper provides a comprehensive review of metal additive manufacturing in the aerospace industry (from industrial/popular as well as technical literature).
[81]
Recent advancement in 3-D printing: nanocomposites with ... - NIH
They can be used to print objects which need to withstand load such as aerospace parts and automobile parts. ... reduce fuel consumption, weight reduction ...
[82]
Samsung Shows Off 600-Mile Solid-State Battery - GreenCars
Aug 27, 2024 · Samsung has just shown off an EV-specific solid-state oxide battery that claims to deliver 600 miles of range, along with all the other benefits of solid-state.Missing: developments | Show results with:developments
[83]
Drone swarms: How they actually work and what industries should ...
Aug 3, 2025 · In the right cases, drone swarms can cut operating costs by up to 30% over time. Most of that comes from lower labor costs, which can make ...
[84]
Case studies: Effective use of Machine learning in manufacturing.
Jul 25, 2025 · Using convolutional neural networks to process high-resolution images has allowed BMW to decrease manufacturing flaws by around 40 percent, a ...
[85]
Machine Learning for Supply Chain Optimization - OptimizePros
Jun 28, 2025 · Just discovered how machine learning cuts supply chain costs by 40% while competitors struggle with outdated methods that drain profits.
[86]
[PDF] Hyperloop Alpha - Tesla
Hyperloop is a new mode of transport that seeks to change this paradigm by being both fast and inexpensive for people and goods. Hyperloop is also unique in ...Missing: AI | Show results with:AI
[87]
Hyperloop Autonomous Systems
Aug 23, 2025 · The development of Hyperloop systems is being accelerated ... AI and machine learning is crucial for developing autonomous control systems.
[88]
Decoherence Time - an overview | ScienceDirect Topics
The decoherence time is more like 10 −3 s in the case of electron spin. It is approximately 10 −6 s for a quantum dot, and around 10 −9 s for an electron in ...
[89]
New quantum computing algorithm skips past time limits imposed by ...
Oct 5, 2020 · "Quantum computers have a limited time to perform calculations before their useful quantum nature, which we call coherence, breaks down," said ...
[90]
Meeting the Challenges of Deep Tech Investing
May 17, 2021 · BCG explores the challenges of investing in digital technology and offers insight into how companies can unlock the power of deep tech.
[91]
Bridging the digital divide: artificial intelligence as a catalyst for ...
85 % of AI health equity studies tracked outcomes for less than 12 months. •. Algorithmic bias leads to 17% lower diagnostic accuracy for minority patients. •.
[92]
Gene editing using CRISPR/Cas9: implications for dual-use ... - NIH
4 Dec 2017 · CRISPR/Cas9 has dual-use risks, including potential for bio-enhancement, bioagent creation, and bioweapon manipulation, and can be used to ...
[93]
The expansionary strategies of intellectual monopolies: Google and ...
Nov 16, 2022 · This paper proposes that these companies are currently entering sectors based on their data-driven intellectual monopoly power.
[94]
[PDF] Accelerating Global Access to Gene Therapies: Case Studies from ...
While clinical research remains limited to HICs, LMICs carry nearly 90% of the global disease burden. 14 In order to develop gene therapies that are safe and ...
[95]
The translational gap for gene therapies in low- and middle-income ...
May 8, 2024 · In April 2024, of the 32 approved gene therapies worldwide, only 5 were granted regulatory approval in countries defined as low- or middle- ...
[96]
Quantum Computing in Finance: Regulatory Readiness, Legal Gaps ...
Sep 9, 2025 · The legal and regulatory analysis adopts a doctrinal framework to dissect statutory and policy responses to quantum computing risks. Primary ...
[97]
EU begins coordinated effort for Member States to switch critical ...
Jun 24, 2025 · EU begins coordinated effort for Member States to switch critical infrastructure to quantum-resistant encryption by 2030.
[98]
Artificial Intelligence and Next Gen Technologies | ENISA
Regulation (EU) 2024/1689 entered into force on 1 August 2024 and aims to support the responsible development and application of AI in the EU. By adopting a ...Missing: encryption | Show results with:encryption
[99]
[PDF] Reflecting on the Five-Year Anniversary of IBM's AI Ethics Board
IBM's AI Ethics Board, formed in 2019, provides governance for AI development, aligning company values with ethics and advancing trustworthy AI.
[100]
Data Hazards: An open-source vocabulary of ethical hazards for ...
The Data Hazards are a vocabulary of 11 ethical hazards in data intensive work. The project provides structure for discussion and reflection to prevent Data ...
[101]
Deep Tech Claims a 20% Share of Venture Capital, Surging Two ...
Nov 21, 2023 · Venture capital funding of deep tech fell from $160 billion in 2021 to about $105 billion in 2022 to $40 billion for the first half of 2023— ...
[102]
The CHIPS Act: What it means for the semiconductor ecosystem - PwC
and whether — to pursue a piece of the $52.7 billion in federal subsidies ...
[103]
Horizon Europe (HORIZON) | EU Funding & Tenders Portal
Horizon Europe is the ambitious EU research & innovation framework programme for 2021-2027 with a budget of €95.5 billion · to strengthen the EU's scientific and ...
[104]
[PDF] Spinning out Success: - Institute for Manufacturing (IfM)
Notably, university spinouts make up a substantial proportion of top-ranking start-ups in terms of investment, comprising 60% of the top 25 pharmaceutical and ...
[105]
Patent Valuation Methods for Deep Tech Intellectual Property
Oct 23, 2025 · Learn about patent valuation methods for deep tech intellectual property and how to maximize the value of your innovations.
[106]
Horizon Europe - Research and innovation - European Commission
With a proposed € 175 billion budget, the programme will boost Europe's productivity and competitiveness, while also improving the well-being of millions of ...Horizon Europe · Horizon Europe programme... · EU Missions in Horizon...
[107]
U.S. Semiconductor Legislation and Policies - Chips Act
The CHIPS Act allocated $52.7 billion to semiconductor manufacturing, research, and workforce development, alongside $24 billion in tax credits to incentivize ...
[108]
Why Billion-Dollar Technology Solutions Die Between Discovery ...
Oct 29, 2025 · Success rate: Only 10%. Traditional funding sources reach ... Overcoming the valley of death: A new model for high technology startups.
[109]
An Entrepreneur's Guide to Surviving the “Death Valley Curve”
Apr 13, 2022 · According to recent estimates, around 90% of start-ups fail. With the global start-up economy valued at $3 trillion, much is at stake.
[110]
Moderna's cash juggernaut rolls on with record $604M IPO
Dec 7, 2018 · Moderna's IPO always looked like a record breaker for biotech and came in even larger than expected at $604 million, valuing the company at an eye-watering $7. ...
[111]
GV - Portfolio
ones where we can provide value across multiple ...Frontier Tech · Life Sciences · AI · Consumer
[112]
The Rise of Thematic Venture Capital Funds: Climate, Deep Tech ...
Oct 3, 2025 · That still amounted to roughly $32 billion globally in 2023, and since 2020, cumulative climate investment has surpassed $140 billion across ...A Market Holding Its Ground · Climate Funds: From Metrics... · Deep Tech And The Long View<|separator|>