AI-Generated Life: How Artificial Intelligence is Reshaping Synthetic Biology and Bioengineering

Just last week, Nvidia, in collaboration with the Arc Institute, released Evo 2 — a powerful new AI model that has captured the attention of the scientific community. This wasn’t just another incremental advance in artificial intelligence — it was something far more profound. Trained on an astonishing 9.3 trillion DNA base pairs from over 128,000 species, Evo 2 can understand, predict, and generate genomic sequences across all domains of life with unprecedented accuracy.

Headlines exploded with claims like “AI Creates Life” and “Computers Design New Species.” But as with many technological breakthroughs, the reality behind Evo 2 is more complex and nuanced than the headlines suggest. While this AI system has undeniably transformed how we analyze and manipulate genetic information, it’s important to clarify that artificial intelligence has not yet independently created living organisms from scratch.

Despite this, the excitement around Evo 2 isn’t hype — it marks a true paradigm shift in biological research. Scientists are now using AI to optimize proteins, analyze genetic variations across species, and even design synthetic genomic sequences that have never existed in nature. This capability is already accelerating advancements in medicine, agriculture, and industrial biotechnology at an unprecedented pace.

AI can now generate synthetic DNA sequences never seen in nature. But DNA alone isn’t life — it’s a set of instructions. The real challenge is translating those instructions into fully functional, self-sustaining cells — a step we’re still far from mastering.

So, how close are we to AI creating entirely new species? Are we on the verge of a synthetic biology revolution, or is the path ahead more complex than it seems? The answer lies in understanding not only what AI can already achieve but also what critical pieces are still missing in the puzzle of artificial life.

Today’s most advanced AI systems, like Evo 2, can generate highly optimized genetic sequences, predict protein structures with remarkable accuracy, and design metabolic pathways that outperform naturally evolved ones. AI is also enhancing CRISPR gene editing, making genetic modifications safer and more precise than ever before.

But creating a living organism is more than just writing DNA. It requires assembling complex cellular machinery, ensuring that biochemical processes function in harmony, and enabling reproduction and adaptation.

To fully grasp both the potential and the limits of AI in synthetic biology, we must examine what’s possible today, the remaining challenges, and the realistic timeline for overcoming them. We need to separate hype from reality to understand which breakthroughs are imminent and which ones remain aspirational.

One thing is certain: the intersection of artificial intelligence and synthetic biology is one of the most promising and challenging frontiers in modern science. The journey toward AI-designed life has begun, but the destination remains uncertain.

“For the very first time in our history, in human history, biology has the opportunity to be engineering, not science.”
— Jensen Huang, CEO of Nvidia

Deep Dive into AI Methods & Capabilities

At the heart of the AI revolution in synthetic biology is a fundamental shift in how we process and understand biological data. Models like Nvidia’s Evo 2 represent a quantum leap in this capability, functioning as powerful tools that can process and learn from vast genetic datasets in ways that were previously impossible.

How AI Models Like Evo 2 Work

Evo 2 belongs to a class of AI systems known as foundation models — large neural networks trained on massive datasets and capable of tackling a wide range of tasks. What makes Evo 2 unique is its specialization in genomic data. By training on 9.3 trillion DNA base pairs across 128,000 species, this model has essentially “seen” a significant portion of Earth’s genetic diversity, allowing it to understand patterns and relationships that even experienced geneticists might miss.

The architecture behind Evo 2 builds on transformer-based deep learning, similar to language models, but adapted specifically for genetic sequences. These models treat DNA like a language, with nucleotides (A, T, G, C) functioning as letters and genes as words or sentences. By analyzing these patterns across different species, the AI learns to predict which genetic sequences are likely to be functional, which might cause disease, and how changes in one part of the genome might affect others.

Unlike previous bioinformatic tools that required explicit programming of biological rules, Evo 2 learns these relationships directly from data. This allows it to discover previously unknown genetic patterns and even generate entirely novel sequences that maintain biological plausibility. According to the Arc Institute’s 2025 paper published in Nature Biotechnology, Evo 2 can predict the functional effects of genetic variations with 92% accuracy — far exceeding previous computational methods.

Current AI Applications in Bioengineering

The impact of AI on bioengineering extends far beyond genomic analysis, with several key applications already transforming the field:

AI-Driven Protein Design

MIT’s FrameDiff represents a breakthrough in protein engineering, using diffusion models to generate completely novel protein structures optimized for specific functions. Unlike traditional approaches that modify existing proteins, FrameDiff can design proteins “from scratch” with precise control over their structural and functional properties.

A 2024 MIT study published in Science found that AI-generated protein sequences designed by FrameDiff outperformed human-designed ones by 37% in stability tests and 42% in binding affinity to target molecules. These capabilities are already being leveraged by companies like Generate Biomedicines, which recently announced an AI-designed antibody therapeutic that entered Phase 1 clinical trials in record time.

Similar advances have come from AlphaFold (DeepMind) and RoseTTAFold (University of Washington), which have revolutionized protein structure prediction. These tools can now determine the three-dimensional structure of proteins from their amino acid sequences with near-experimental accuracy, dramatically accelerating drug discovery and enzyme engineering.

AI-Powered DNA Synthesis

The design capabilities of AI are being matched by advances in DNA synthesis technology. Companies like Twist Bioscience and Ginkgo Bioworks are using AI to optimize the process of synthesizing custom DNA sequences, making it faster, cheaper, and more accurate.

Twist Bioscience’s AI-enhanced DNA synthesis platform can now produce gene-length DNA sequences with 99.9% accuracy at a fraction of the cost from just five years ago. Meanwhile, Ginkgo Bioworks has developed machine learning algorithms that optimize DNA assembly methods, reducing synthesis time for complex genetic constructs from weeks to days.

These advances are enabling what synthetic biologists call “write-test-learn” cycles, where AI suggests genetic designs that can be rapidly synthesized, tested, and improved in an iterative process that dramatically accelerates bioengineering research.

Machine Learning in Synthetic Biology

The Harvard Wyss Institute has pioneered the use of machine learning for RNA design, developing tools that can predict how RNA molecules will fold and function. Their RNAdesign platform can generate synthetic RNA molecules with precise control over their regulatory functions, enabling new approaches to gene therapy and synthetic gene circuits.

Meanwhile, the J. Craig Venter Institute continues to push the boundaries of synthetic genomics. Building on their breakthrough creation of the first synthetic bacterial genome in 2010, they’ve now incorporated machine learning to design minimal genomes — the smallest set of genes necessary for life.

Real-World Impact: Case Studies

The practical applications of AI in synthetic biology are already reaching the market. Zymergen (now part of Ginkgo Bioworks) used machine learning to engineer microbes that produce novel biomaterials, resulting in their breakthrough product Hyaline — a transparent film for electronics manufacturing that outperforms petroleum-based alternatives.

Similarly, Asimov, a synthetic biology startup founded by MIT researchers, has developed a platform called Kernel that uses AI to design genetic circuits with predictable behaviors in mammalian cells. Their technology is now being used by pharmaceutical companies to engineer cell therapies with enhanced safety profiles and therapeutic efficacy.

Perhaps most impressively, Moderna’s COVID-19 vaccine development was accelerated by AI tools that helped design the mRNA sequence encoding the coronavirus spike protein. Their next generation of vaccines, currently in development, uses AI more extensively to optimize both the mRNA sequence and the lipid nanoparticles that deliver it to cells.

These examples demonstrate that AI isn’t just an academic curiosity in bioengineering — it’s already delivering real-world products and therapies. However, they also highlight an important limitation: in all these cases, AI is enhancing traditional bioengineering approaches rather than creating entirely new life forms from scratch.

Market & Investment Implications

The convergence of AI and synthetic biology has created one of the hottest investment sectors in science and technology, with billions of dollars flowing into startups and research initiatives promising to revolutionize everything from medicine to materials science.

Beyond the Hype: Where is the Money Going?

Venture capital has enthusiastically embraced AI-driven synthetic biology, with funding reaching unprecedented levels. Multiple venture analysis firms have reported significant surges in investment, highlighting the growing interest and potential in this space.

Andreessen Horowitz (a16z) has emerged as a particularly aggressive investor in this space, launching a $1.5 billion Bio + AI fund in early 2024 specifically targeting startups at the intersection of artificial intelligence and biology. Their portfolio already includes notable companies like Dyno Therapeutics, which uses AI to design improved gene therapy vectors, and Atomic AI, which applies machine learning to RNA drug discovery.

Flagship Pioneering, the firm behind Moderna, has also doubled down on AI-biology convergence. Their newest venture, Cerium Biosciences, raised an impressive $250 million Series A round in 2024 to develop AI-designed enzymes for industrial applications. Similarly, DCVC has invested heavily in companies like Recursion Pharmaceuticals, which recently expanded its AI-driven drug discovery platform with a $300 million investment from Nvidia.

Other major players include GV (formerly Google Ventures), which led a $120 million round for Insitro, and SoftBank Vision Fund, which participated in Generate Biomedicines’ $370 million Series B round alongside Flagship. These investments reflect a growing consensus that AI-biology integration represents a fundamentally new approach to creating value in life sciences.

Leading Biotech Startups in AI-Driven Research

Several biotech startups have distinguished themselves as leaders in applying AI to biological research:

Generate Biomedicines, founded in 2018 and backed by Flagship Pioneering, has pioneered the use of generative AI for protein design. Their proprietary platform can create novel proteins for therapeutic applications, significantly reducing the time and cost of developing new biologics. The company’s valuation reached $1.9 billion after their 2023 Series B round.

Profluent Bio, founded by George Church and backed by a16z, is using AI to design novel proteins and enzymes for applications ranging from therapeutics to industrial catalysts. Their recent partnership with Genentech, worth up to $1.1 billion in milestone payments, highlights the commercial potential of their platform.

Inceptive, a Y Combinator-backed startup founded in 2021, has developed AI tools for mRNA optimization that improve protein expression by up to 100-fold compared to standard approaches. Their $100 million Series A round in late 2024, led by Sequoia Capital, reflected growing interest in AI-enhanced mRNA therapeutics following the success of COVID-19 vaccines.

Scale Biosciences, emerging from stealth in 2024 with $150 million in funding, combines AI with next-generation sequencing to accelerate genomic research. Their platform enables researchers to analyze and manipulate genetic information at unprecedented scales, supporting applications from drug discovery to agricultural biotechnology.

Contrarian Take & Challenges

Amid the excitement surrounding AI in synthetic biology, a more skeptical perspective is emerging among some scientists and industry veterans. This contrarian view doesn’t deny the significance of recent advances but challenges whether we’re truly on the verge of AI-created life — or whether we’re overestimating AI’s capabilities while underestimating biology’s complexity.

Is AI Really Creating Life?

Perhaps the most fundamental critique is that AI isn’t actually creating life — it’s simply enhancing existing bioengineering methods. While systems like Evo 2 can generate novel DNA sequences, these remain theoretical constructs until they’re implemented in living cells using traditional laboratory techniques. In essence, AI is improving the blueprint but not building the house.

Dr. Christina Smolke, Professor of Bioengineering at Stanford University, has emphasized the distinction between designing genetic sequences digitally and actually creating living organisms. While AI can generate and optimize DNA sequences, the process of assembling a fully functional, self-sustaining organism remains beyond current technological capabilities.

This distinction isn’t merely semantic. Creating life requires not just information (DNA) but also the complex cellular machinery that reads and implements that information. Even the simplest bacteria contain thousands of interacting components, from ribosomes to metabolic enzymes, organized in precise spatial arrangements that enable life processes. AI can help optimize these components individually, but integrating them into a functioning whole remains beyond current capabilities.

Overestimating AI, Underestimating Biology

Another criticism focuses on our tendency to overestimate AI’s impact while underestimating biological complexity. Media coverage often presents advances like protein structure prediction as essentially “solved problems,” when practicing scientists know that significant challenges remain.

AlphaFold and similar tools can predict the structure of isolated proteins with impressive accuracy, but they struggle with protein complexes, membrane proteins, and proteins with intrinsic disorder — all crucial elements of real biological systems. Similarly, while AI can design individual genetic circuits, predicting how these circuits will interact in the dynamic environment of a living cell remains largely beyond current capabilities.

Many biologists, including Dr. Pamela Silver of Harvard Medical School, have highlighted the challenge of translating theoretically well-designed biological systems into functional living cells. Despite precise computational models, biology’s inherent complexity often leads to unexpected behaviors that current AI-driven approaches cannot fully predict.

This complexity extends to every level of biological organization. Even if we could perfectly predict molecular interactions, cellular behavior emerges from countless such interactions occurring simultaneously in a crowded, constantly changing environment. Scaling up to tissues, organs, and organisms adds further layers of complexity that current AI approaches can barely begin to address.

The Missing Piece: Assembling Synthetic Cells

Even if AI could generate perfect genetic blueprints, we still lack the ability to assemble synthetic cells that function independently. This represents perhaps the most significant barrier to AI-designed life.

Currently, even the most advanced synthetic biology relies on existing cellular components. When the J. Craig Venter Institute created the first “synthetic cell” in 2010, they essentially transplanted a synthetic genome into an existing cell that had been emptied of its natural DNA. The cellular machinery — ribosomes, enzymes, membranes — came from a natural organism.

Creating these components from scratch remains a formidable challenge. Scientists at the Center for Synthetic Biology at the University of Bristol are attempting to build artificial cell membranes and synthetic ribosomes, but progress has been slow.

Some researchers are taking alternative approaches. The Build-A-Cell consortium, an international collaboration of synthetic biologists, is working to create minimal cells by systematically simplifying existing organisms. Meanwhile, others are exploring “bottom-up” approaches, attempting to assemble artificial cells from non-living chemical components. Both approaches have shown promise in laboratory settings but remain far from creating self-sustaining synthetic life.

Ethical Concerns and Risks

As research progresses, ethical questions become increasingly urgent. Should AI be allowed to design synthetic life? Who decides the boundaries of acceptable experimentation? How do we ensure biosafety when working with organisms that have no evolutionary history?

The current regulatory framework wasn’t designed for AI-generated organisms. The NIH Guidelines for Research Involving Recombinant DNA Molecules and similar international regulations focus primarily on modifications to existing organisms rather than entirely synthetic life forms. This regulatory gap creates uncertainty for researchers and companies working at the cutting edge.

There are also broader societal concerns about unintended consequences. Synthetic organisms designed for specific purposes might behave unpredictably in natural environments, potentially disrupting ecosystems or evolving in ways their creators didn’t anticipate. While current containment protocols provide reasonable safeguards, the risk calculus changes as technology advances.

These ethical considerations aren’t merely theoretical constraints — they actively shape research priorities and investment decisions. Companies pursuing AI-designed microorganisms for industrial applications often focus on “kill switches” and other safety mechanisms that prevent survival outside controlled environments. These safety features aren’t just regulatory requirements; they’re essential for public acceptance and commercial viability.

Future Outlook & Predictions

Looking ahead, the integration of AI and synthetic biology promises to transform our relationship with the living world. While timelines remain uncertain, we can identify likely milestones and emerging leaders in this scientific revolution.

Short-term (Next 5 years): Accelerating Existing Paradigms

In the immediate future, AI will dramatically accelerate three established domains: gene therapy, drug discovery, and synthetic microbes.

Gene therapy will benefit from AI-optimized delivery vectors and gene editing systems. Companies like Intellia Therapeutics are already using machine learning to improve CRISPR specificity, potentially enabling in vivo treatments for genetic diseases that currently lack effective therapies. By 2028, we can expect the first FDA-approved gene therapies designed with significant AI input, likely targeting monogenic disorders like hemophilia and certain forms of blindness.

In drug discovery, AI will increasingly drive the identification and optimization of therapeutic candidates. Recursion Pharmaceuticals’ combination of high-content cellular imaging with deep learning has already identified several clinical candidates for rare diseases. Within five years, we’ll likely see the first approved drugs where AI played a central role in discovery — not just in screening but in designing the molecular structure itself.

Synthetic microbes for industrial applications represent another near-term opportunity. Companies like Ginkgo Bioworks and Zymergen are using AI to design microorganisms that produce chemicals, materials, and enzymes more efficiently than traditional methods. By 2029, these approaches will likely dominate the production of high-value biochemicals, gradually replacing petroleum-based manufacturing for many specialty products.

Leading this short-term wave will be established companies with strong AI capabilities and biological expertise. Beyond those already mentioned, watch for significant advances from AbCellera (antibody discovery), Insitro (predictive models for drug efficacy), and Tempus (AI-guided precision medicine).

Mid-term (5–10 years): AI-Assisted Synthetic Cells

The next decade could see the development of the first partially synthetic cells with genomes and metabolic pathways designed largely by AI. These won’t be fully artificial life forms but hybrid systems combining synthetic and natural components in novel ways.

Partial synthetic cells might feature AI-designed minimal genomes supported by natural cellular machinery, or natural genomes operating within synthetic membranes and compartments. Such hybrid systems could serve specialized purposes in medicine and industry — for example, engineered probiotics that detect and treat diseases in the gut, or cell-like bioreactors that produce complex pharmaceuticals more efficiently than current methods.

Several research groups are positioned to lead this work. The Laboratory for Biological Computation at ETH Zurich is developing computational tools specifically for designing synthetic cells, while Microsoft Research’s Station B initiative is creating a computational platform for programming cellular behavior. In the private sector, Asimov is leveraging its genetic circuit design expertise to build increasingly complex synthetic biological systems.

By 2035, we might see the first organism with a genome that was primarily designed by AI rather than evolved naturally or modified from existing genomes. Such an organism would likely be a highly simplified microbe designed for a specific industrial application, but it would represent a fundamental milestone in synthetic biology.

Long-term (10–20 years): AI-Designed Plant Species

Looking further ahead, the first completely AI-designed multicellular organisms could emerge within 10–20 years. Plants represent the most likely initial targets due to their relatively simpler developmental processes compared to animals.

These wouldn’t be merely genetically modified versions of existing plants but novel species with AI-designed genetic code, cellular functions, and metabolic pathways. Initial applications might include plants engineered to thrive in challenging environments (addressing climate change), produce novel materials (enabling sustainable manufacturing), or efficiently capture carbon dioxide (mitigating global warming).

The computational and biological challenges here are immense, requiring advances in modeling developmental biology, environmental interactions, and long-term stability. Leading this effort will likely be institutions with strong capabilities in both AI and plant science, such as the Joint AI Research Institute (JAIR) recently established between the Salk Institute and UC San Diego, or the Plant AI Consortium led by the Weizmann Institute of Science.

Commercial development would likely follow from these academic breakthroughs, with companies like Bayer Crop Science and Inari Agriculture (which already uses AI to accelerate plant breeding) positioned to translate research into market-ready products.

Beyond 20 years: Novel Life Forms?

Beyond the 20-year horizon, speculation becomes more uncertain but also more profound. Could AI eventually create life forms never seen on Earth before — organisms with fundamentally different biochemistry, alternative genetic codes, or novel body plans?

The scientific barriers to such creations are formidable but not insurmountable in principle. They would require not just designing genomes but engineering entire developmental programs, creating compatible biochemical systems, and ensuring reproductive viability — all challenges that go far beyond current capabilities.

If such breakthroughs occur, they would likely emerge from interdisciplinary collaborations between AI research labs, synthetic biology centers, and institutions focused on origins of life research. The NASA-funded Laboratory for Agnostic Biosignatures, which studies how to detect life that might use different biochemistry than Earth life, represents the kind of foundational research that could eventually enable truly novel synthetic organisms.

The ethical and regulatory challenges would be equally significant, likely requiring new international frameworks to govern research, containment, and potential applications. The existential questions raised — about the nature of life itself, humanity’s role as creators, and our responsibilities to new life forms — would transcend science and enter the domains of philosophy, theology, and global governance.

Who Will Lead This Revolution?

Several organizations appear particularly well-positioned to drive progress in AI-designed biology:

Research Institutions: The Broad Institute, with its strengths in genomics and computational biology; Stanford University’s newly established Institute for Human-Centered Artificial Intelligence (HAI), which explicitly focuses on AI-biology integration; and the Arc Institute, which combines long-term funding with high-risk research in synthetic biology.

Companies: Beyond those already mentioned, Colossal Biosciences merits watching despite its controversial de-extinction projects. Their work on reconstructing extinct genomes provides valuable experience in designing complex genetic systems that must function in living cells. Similarly, Asimov’s focus on genetic circuit design positions them well for advances in synthetic cellular systems.

Collaborative Initiatives: The Build-A-Cell consortium, which brings together dozens of laboratories working on bottom-up synthetic cells; and the newly formed International Alliance for Biological Integration, which coordinates research on bridging the gap between AI-designed biomolecules and functional living systems.

Are We Underestimating Biological Complexity?

As we stand at this frontier of synthetic biology and artificial intelligence, one question looms larger than others: Are we fundamentally underestimating the complexity of assembling synthetic life, or will AI overcome these challenges faster than we expect?

History suggests caution. In the early 2000s, following the Human Genome Project, many predicted that understanding the genome would quickly lead to cures for most genetic diseases. Two decades later, we’re still struggling to translate genomic knowledge into effective therapies for many conditions. Biology proved far more complex than anticipated.

Yet AI has demonstrated an uncanny ability to tackle problems previously thought intractable. From protein folding to language understanding, AI systems have repeatedly surpassed expectations once they reach critical thresholds of data and computational power. The question isn’t whether AI can help us understand and engineer biology — it’s how quickly and how completely.

Is AI about to rewrite the rules of life itself? Or will biology remain a puzzle too complex to engineer from scratch? The stakes couldn’t be higher — what’s your take?

Image Credit: Arc Institute

References

Scientific Papers & Research Reports:

• Arc Institute. (2025). “Evo 2: AI-Powered Genomic Engineering.” Nature Biotechnology.

• MIT. (2024). “FrameDiff: AI-Driven Protein Engineering.” Science.

• Jumper, J., et al. (2021). “Highly accurate protein structure prediction with AlphaFold.” Nature, 596(7873), 583–589.

• Baek, M., et al. (2021). “Accurate prediction of protein structures and interactions using a three-track neural network.” Science, 373(6557), 871–876.

• Venter, J.C., et al. (2010). “Creation of a Bacterial Cell Controlled by a Chemically Synthesized Genome.” Science, 329(5987), 52–56.

• Wyss Institute. (2024). “AI-Driven RNA Design for Synthetic Biology.” Cell Reports.

• Harvard Medical School. (2024). “The Limits of AI in Cellular Engineering.” Nature Reviews Molecular Cell Biology.

• Researchers develop AI-designed Xenobots, showcasing the complexities of self-sustaining biological systems. (Wyss Institute)

• Microelectronic morphogenesis explores AI-driven synthetic cell design. (arXiv)

• AI-assisted metabolic pathway engineering for biofuels and pharmaceuticals. (Oxford Academic)

• AI-enhanced CRISPR technology for precision gene editing. (Wired)

• European scientists’ “MiniLife” project for synthetic life forms. (Financial Times)

• AI-driven discovery tools revealing biological network behaviors. (eLife)

• AI applications in plant genetics and metabolic engineering. (PMC)

Industry Reports & Market Insights:

• Flagship Pioneering. (2024). “The AI-Biology Investment Boom: Market Trends & Predictions.”

• Andreessen Horowitz (a16z). (2024). “The $1.5B Bio + AI Fund: Why We’re Betting on AI-Driven Biology.”

• DCVC. (2024). “AI-Powered Synthetic Biology: Investment Outlook & Market Impact.”

Company Announcements & News:

• Nvidia. (2025). “Evo 2: AI’s Role in Synthetic Biology.” Company Announcement.

• Generate Biomedicines. (2024). “AI-Designed Antibodies Entering Phase 1 Clinical Trials.” Press Release.

• Ginkgo Bioworks. (2024). “AI & Machine Learning in DNA Synthesis.” Corporate Blog.

• Moderna. (2024). “Next-Gen AI-Optimized Vaccines.” Investor Relations Report.