Synthetic Biology: what are the biggest unsolved problems?
In our synthetic biology market deck, you will find everything you need to understand the market
SUMMARY
Synthetic biology’s biggest unsolved problem is predictable engineering: researchers can edit and build DNA with remarkable precision, but they still cannot reliably predict how the resulting biology will behave across cells, environments, generations and industrial scale.
The gap between DNA engineering and biological engineering is now hard to miss. Sequencing, synthesis, multiplex editing and automated experimentation have improved much faster than our ability to predict phenotype from genotype.
Minimal cells expose how deep the problem goes. Even after stripping a cell down to fewer than 500 genes, researchers have repeatedly discovered essential functions and interactions only after something behaved unexpectedly.
Genome writing is starting to reverse the traditional bottleneck. Building very large pieces of DNA is increasingly possible; deciding exactly what should be written, and getting that design right without long debugging cycles, is becoming the harder part.
Evolution means the finished sequence is never really a permanent specification. If an engineered function costs the cell energy, mutants that weaken or discard that function can gradually take over the population.
Laboratory success also travels badly. Temperature changes, nutrient gradients, competition, immune responses and other real-world conditions can expose weaknesses that remain invisible in controlled experiments.
Industrial scale-up is where biological uncertainty turns directly into economics. A strain can look excellent in a small vessel and lose enough titer, rate or yield inside a large fermenter to make the commercial process unattractive.
AI is changing the speed of synthetic biology faster than it is changing predictability. Autonomous laboratories can now search enormous experimental spaces, which helps engineers find working solutions even when the underlying biological model remains incomplete.
Medicine shows that sophisticated biological programs are becoming clinically credible, especially when cells can be engineered outside the body. Delivering complex genetic logic directly to the right cells inside a patient remains much harder.
Environmental applications push uncertainty even further. Gene drives may work extremely well in cages, but population structure, migration, resistance and ecological interactions make real-world behavior far harder to forecast.
Biosecurity is becoming part of the engineering problem too. As DNA synthesis gets cheaper and biological design gets easier, sequence screening and access controls have to improve without creating so many false alarms that legitimate research becomes cumbersome.
Most of the field’s hardest problems therefore collapse into the same question: can synthetic biology move from fast experimental search toward biology that can be designed, built and trusted much more like an engineered system?
This market map, featured in our synthetic biology market deck, highlights top companies and startups in the synthetic biology market
Why is synthetic biology still so hard when DNA editing has become so good?
Synthetic biology is currently much better at changing DNA than at predicting what those changes will make a living system do.
Researchers can sequence genomes cheaply, order custom DNA, edit several genomic sites at once, generate huge libraries of variants and automate thousands of experiments. Some of those capabilities would have looked extraordinary 15 years ago.
Yet synthetic biology promises something harder than DNA manipulation. We want to start with a desired biological behavior, design the DNA that should produce it, build the system and get something close to the expected result.
We still miss that last step surprisingly often.
NIST’s Engineering Biology program, updated earlier this year, still puts predictive design, comparable measurements and real-world scalability among its main priorities. The US National Security Commission on Emerging Biotechnology reached a similar conclusion in its final report, calling for a multibillion-dollar research challenge to make biology “predictably engineerable,” alongside a separate challenge for predictable and cost-competitive biomanufacturing scale-up.
The trouble starts with context. Genetic parts interact with the host that carries them. They compete for ribosomes, RNA polymerases, metabolites and energy. Changing expression in one part of a synthetic circuit can alter another part indirectly, slow growth or push the cell into a different physiological state.
Recent work on synthetic circuits increasingly models cellular burden, growth feedback and resource competition because engineers kept seeing apparently modular components behave differently after they were combined.
The same issue appears across organisms. Bacterial circuits can fail when growth conditions change. Mammalian circuits face additional layers of chromatin regulation, signaling and cell-state variability. Moving the same design to another strain or cell type can change the output again.
Host-aware models, feedback controllers and high-throughput testing can catch failures earlier. Still, much of synthetic biology works through accelerated search: generate candidates, build them, test them and see which ones survive contact with the cell.
That leaves synthetic biology in a slightly awkward place. We can increasingly build the DNA we want, while reliable prediction still tends to arrive after the experiment rather than before it.
If you want more recent data on this point, please see our latest synthetic biology market report.
Do we actually understand cells well enough to design one from scratch?
Synthetic biology still lacks a complete enough understanding of even a minimal cell to design living systems confidently from first principles.
JCVI-syn3.0 remains one of the cleanest demonstrations. The J. Craig Venter Institute reduced a Mycoplasma genome to 473 genes, creating a cell close to the minimum needed for autonomous growth in laboratory conditions.
When the cell was first reported, researchers could assign no specific biological function to 149 of those genes. Almost one-third of the genome inside one of the simplest independently growing cells on Earth remained functionally unclear.
Researchers have since narrowed that gap considerably. Later JCVI work reduced the number of poorly understood genes, and JCVI-syn3A now provides one of the best-characterized platforms available for whole-cell biology.
Even basic cell division produced surprises. The highly minimized cells divided into strange shapes. Restoring seven genes brought much more normal division back. Several of those genes had unclear functions when researchers discovered their role.
Now the field has gone further. A major whole-cell modeling effort reported this year simulated an entire cell cycle of JCVI-syn3A in four dimensions, combining information on metabolism, gene expression, chromosome organization and molecular structure.
JCVI-syn3A still contains only about 543,000 base pairs and 493 genes, roughly one-tenth the genomic scale of E. coli. Even at that stripped-down scale, we still cannot reliably predict how a newly engineered cell will behave across conditions.
As this chart shows, and as featured in our synthetic biology market deck, search interest in gene editing has grown significantly
Can synthetic biology write whole genomes without years of debugging?
Synthetic biology can now build genome-scale DNA, but designing a large genome that works correctly on the first try remains out of reach.
The Sc2.0 synthetic yeast project is the clearest example. The international consortium redesigned and constructed synthetic versions of all 16 Saccharomyces cerevisiae chromosomes, covering a genome of roughly 12 million base pairs.
That is an enormous engineering achievement. It also took a global collaboration many years, and the project produced a long catalogue of failures that had to be hunted down and corrected.
Synthetic chromosome XVI shows what that debugging looks like. The chromosome contains about 903,000 base pairs. Researchers found that some engineered loxPsym sites interfered with nearby gene regulation, producing growth defects under conditions including glycerol growth and higher temperature. They eventually traced the defects to particular design choices and rebuilt those regions.
A Nature Biotechnology analysis of Sc2.0 published recently focused largely on what went wrong during construction of the 16 chromosomes and how those failures should change future synthetic-genome projects.
Mammalian genome writing is advancing too. Researchers have recently assembled megabase-scale synthetic human DNA and delivered it into mouse embryos. Another recent system called INSTALL improved kilobase-scale writing in mammalian cells using DNA donors designed to provoke much weaker innate immune responses.
A human genome contains around three billion base pairs, hundreds of times more DNA than the yeast genome, with repetitive sequences, long-range regulation, chromatin and epigenetic control layered on top.
The bottleneck is increasingly about knowing what to write, rather than simply proving that very large pieces of DNA can be made.
| Biological scale | What researchers can do now | What still breaks |
|---|---|---|
| Minimal bacterial genome | Build a complete synthetic genome supporting autonomous growth | Gene functions and genome-wide interactions remain partly unclear |
| Yeast genome, ~12 Mb | Construct synthetic versions of all 16 chromosomes | Consolidation, hidden interactions and debugging remain demanding |
| Mammalian DNA, kb to Mb scale | Assemble, deliver and replace increasingly large DNA regions | Delivery, epigenetics, repeats and phenotype prediction remain hard |
| Human genome, ~3 Gb | Make precise targeted edits routinely | Rational whole-genome design remains far beyond current practice |
If you want more recent data on this point, please see our latest synthetic biology market report.
Can engineered cells keep doing the same job after hundreds of generations?
Evolutionary stability remains a major synthetic biology problem because engineered cells can gradually escape the job we designed them to perform.
A production strain may spend energy making an enzyme, protein or chemical that helps us while giving the cell little benefit. A mutant that reduces that expensive activity can grow slightly faster.
Give that mutant enough generations and natural selection starts working against the engineer.
Researchers see the same problem in complex synthetic circuits. Mutations can disable burdensome components, change regulatory sequences or alter copy number. Even without a dramatic mutation, populations can drift toward cells that reproduce better while producing less of the desired output.
The minimal-cell experiments show how quickly biology can find its own route. Genome minimization initially cut fitness by more than half in one experiment. After 2,000 generations of evolution, the minimal cells recovered the lost fitness.
Industrial production makes small advantages add up. If a low-producing mutant grows only slightly faster than the intended production strain, repeated generations inside a long fermentation can change the population.
Genome integration, lower-burden circuits, kill switches, addiction systems and growth-coupled production can reduce that risk. A general way to keep complex engineered functions stable for hundreds or thousands of generations still does not exist.
This chart, featured in our synthetic biology market deck, illustrates yearly VC funding for synthetic biology startups
Can synthetic biology work outside perfect laboratory conditions?
Synthetic biology still struggles to make engineered organisms behave consistently once temperature, nutrients, competitors and other environmental conditions start moving around.
Laboratories deliberately remove much of that mess. Researchers choose the medium, temperature, oxygen level, growth phase and strain. Real applications rarely offer that level of control.
Sc2.0 exposed the issue at genome scale. Some synthetic yeast defects appeared only under particular conditions, including alternative carbon sources and elevated temperature. A genome could therefore look healthy during ordinary testing and reveal weaknesses only when the environment changed.
The problem becomes harsher in soil, the gut, industrial waste streams or open ecosystems. Engineered organisms encounter fluctuating nutrients, other species, immune responses, toxins and physical stresses that are hard to recreate fully during development.
This helps explain the current push toward non-model organisms. E. coli and Saccharomyces cerevisiae are popular partly because researchers have spent decades building excellent tools for them. The organism best understood in the laboratory may be a mediocre organism for the real job.
A recent Nature Reviews Bioengineering analysis makes that case explicitly. Native producers such as pseudomonads and Lacticaseibacillus can start with metabolic abilities that engineers would otherwise have to transplant into E. coli. Other organisms naturally tolerate high salt, unusual feedstocks, toxic compounds or extreme temperatures.
Many of those useful organisms remain difficult to transform, poorly characterized or short on standardized promoters, vectors and genome-editing tools. That still pushes researchers toward organisms that are easier to engineer even when another organism might be better suited to the application.
Can synthetic biology scale cheaply enough to beat petrochemicals?
Industrial synthetic biology currently has two linked problems: great strains often lose performance in large fermenters, and even successful scale-up may still be too expensive for commodity chemicals.
A small fermentation vessel mixes quickly. Oxygen, nutrients and temperature can stay fairly uniform. Once the reactor reaches thousands of litres, cells experience a much more uneven world.
A major Nature Communications review published this year describes nutrient and oxygen gradients as a core reason microbial production loses performance during scale-up. Cells circulate through different microenvironments inside the same tank and repeatedly switch metabolic state.
Scale changes mass transfer, mixing, shear, heat removal and the time cells spend under nutrient or oxygen limitation. Enough deterioration can erase years of strain optimization.
Some recent work shows that the barrier can be beaten for particular products. An engineered Corynebacterium glutamicum process reported this year reached 141.5 grams per litre of 1,3-propanediol at 2.95 grams per litre per hour without antibiotic selection, after the researchers transferred the design into a newly isolated production strain.
Then the economics get brutal.
A specialty pharmaceutical ingredient can tolerate high manufacturing costs because the final product may sell for hundreds or thousands of dollars per gram. A fuel or bulk chemical may compete in markets where every extra dollar per kilogram matters.
Three biological numbers dominate those economics: titer, rate and yield. Titer tells us how concentrated the product becomes. Rate tells us how quickly the reactor makes it. Yield tells us how much useful product comes from the feedstock.
A Lawrence Berkeley National Laboratory analysis published this year modeled several biofuel pathways and found strongly nonlinear economics. For the systems studied, commercially attractive low-carbon production required near-theoretical yield, titers above roughly 130 grams per litre and production rates above roughly 2 grams per litre per hour.
Those thresholds vary by product, but their order of magnitude is useful. A strain producing a few grams per litre can be scientifically impressive and economically miles away from a commodity process.
Downstream processing can make the gap worse. A microorganism may produce a molecule efficiently while extraction and purification remain expensive.
This is why high-value products such as enzymes, pharmaceutical ingredients, flavors and specialized materials reached commercial synthetic biology earlier. Commodity markets demand excellent fermentation, scale-up and purification at the same time.
| Economic variable | What it changes in the real process |
|---|---|
| Titer | Low concentration means more fermentation volume and more material to separate |
| Rate | Slow production ties up expensive reactor capacity |
| Yield | Poor conversion wastes feedstock and raises raw-material costs |
| Scale robustness | Great laboratory numbers lose value if they collapse in a large tank |
| Recovery | Difficult purification can dominate the final cost even when fermentation works well |
If you want more recent data on this point, please see our latest synthetic biology market report.
This chart, featured in our synthetic biology market deck, shows how Twist Bioscience is capturing share in synthetic biology
Are AI and autonomous labs finally making synthetic biology predictable?
AI and autonomous labs are making synthetic biology dramatically faster today, but their strongest results still come from running huge numbers of experiments rather than perfectly predicting biology in advance.
The most striking recent example came from OpenAI and Ginkgo Bioworks.
GPT-5 was connected to Ginkgo’s automated laboratory and given the task of optimizing cell-free protein synthesis. Across six experimental rounds, the system tested more than 36,000 reaction compositions on 580 automated plates. It eventually cut total protein-production cost by 40% against the previous benchmark while increasing yield.
A model can now propose thousands of conditions, robots can run them, the results can flow back automatically and another experimental round can begin with little human intervention.
Three rounds were enough to establish the new low-cost benchmark once the model had the required tools and literature.
Yet GPT-5 had access to physical feedback at enormous scale. The laboratory kept telling the model which ideas worked.
Protein-design models show a similar pattern. AI can generate plausible structures and sequences at extraordinary speed, while experiments still determine expression, activity, toxicity, specificity and behavior in the real biological environment.
Whole-cell modeling remains much harder. As seen above, one of the most advanced current simulations focuses on a minimal bacterium with fewer than 500 genes.
Automation also needs common measurements and protocols if results are going to transfer cleanly between facilities. NIST continues to work on engineering-biology standards, reference measurements and reproducibility, while biofoundries are developing more interoperable descriptions of workflows and operations.
AI can already compensate for weak prediction by making experimentation vastly faster. That is a real change in how synthetic biology gets done, even if the underlying biology is still not fully predictable.
Can synthetic gene circuits actually become routine medicines?
Synthetic gene circuits are finally producing credible clinical results, but routine programmable medicine still depends heavily on whether we can deliver those circuits safely and make them behave consistently inside patients.
Senti Biosciences gives us one of the best current examples.
SENTI-202 is an engineered natural-killer-cell therapy for acute myeloid leukemia. Its genetic logic circuit targets cells expressing CD33 or FLT3 while using a separate inhibitory signal designed to help protect healthy hematopoietic cells.
The company completed enrollment in its Phase 1 study earlier this year after presenting deep measurable-residual-disease-negative complete remissions in treated patients. SENTI-202 has also received the FDA’s Regenerative Medicine Advanced Therapy designation.
That takes synthetic cellular logic well beyond a laboratory demonstration.
The wider medical field is moving quickly too. The FDA said early this year that its center had approved close to 50 cell and gene therapies over the previous decade. Since then, additional engineered therapies have reached the market, including a new genetically modified oncolytic viral therapy for advanced melanoma.
Still, programmable circuits remain a small corner of that clinical success. A Cell Systems review published recently found that clinical translation has concentrated heavily on simpler circuit designs, with relatively few sophisticated systems reaching human trials.
Delivery is part of the reason.
A genetic circuit has to survive administration, reach the correct tissue, enter the intended cell type, release its cargo and avoid an immune reaction strong enough to stop treatment. Viral vectors handle parts of that chain very well, although cargo capacity, immunity and repeat dosing can become limiting. Lipid nanoparticles have become exceptionally useful for RNA and liver delivery, while many other tissues remain much harder to target precisely.
Recent genome-writing work shows how actively researchers are attacking the problem. The INSTALL system reported this year uses circular single-stranded DNA donors designed to reduce innate immune activation and improve tolerability during large-sequence integration. Researchers demonstrated the approach in primary human cells and in mice.
Ex-vivo cell therapy avoids part of the delivery problem by engineering cells outside the body and infusing them back into the patient. The process gives researchers more control, although manufacturing becomes cumbersome.
SENTI-202 makes programmable medicine much more credible than it was a few years ago. Directly installing complex genetic programs into precisely chosen cells inside the body remains a much harder target.
If you want more recent data on this point, please see our latest synthetic biology market report.
This chart, featured in our synthetic biology market deck, illustrates yearly funding for synthetic biology startups
Can synthetic biology safely release gene drives into the wild?
Synthetic biology has yet to prove that self-spreading gene drives can be released into natural ecosystems with predictable enough behavior for routine use.
A gene drive biases inheritance so a chosen genetic trait can spread through a population much faster than ordinary Mendelian inheritance would allow.
For malaria, the potential benefit is enormous. Researchers can design Anopheles mosquitoes that spread infertility or parasite-blocking traits through a population.
Laboratory performance has become impressive. A gene drive targeting the doublesex gene has produced complete suppression in cage populations. Researchers this year also reported multiplexed designs that attack several conserved target sites, aiming to make resistance much harder to evolve.
Yet real mosquito populations are many orders of magnitude larger than laboratory cages. A resistance mutation rare enough to escape detection in a lab population can become relevant when millions or billions of mosquitoes are involved.
A major review published this year still reports that no gene-drive strategy has undergone a field trial. Researchers are now preparing the ecological data, monitoring plans and regulatory work needed for those first releases. WHO documents have described candidate trials approaching that stage.
More conventional self-limiting mosquitoes have already been released outdoors at huge scale. Gene drives create a different risk profile because successful inheritance is designed to keep spreading after the initial release.
Mosquito movement, mating structure, seasonal population changes, predators, competition and migration can all affect the outcome.
As of now, the crucial evidence is still missing: how a true gene drive behaves after release into a real ecosystem.
Can DNA synthesis get much cheaper without making biosecurity worse?
Synthetic biology still lacks a fully mature way to make DNA synthesis widely accessible while reliably screening out sequences that could create serious biological risk.
Cheaper synthesis gives researchers faster experiments, easier genome engineering and broader access to biotechnology. Benchtop synthesis could eventually remove days or weeks from the process of turning a digital sequence into physical DNA.
The same reduction in friction also makes concerning sequences easier to turn into physical material.
NIST began monthly baseline testing of screening systems last year. Its latest published results, covering provider tests through July, show a median sensitivity of 96.75% and median accuracy of 97.88%. Both clear NIST’s current passing thresholds.
Those are encouraging numbers, especially because DNA-order screening was far less measurable only a few years ago.
The harder tests are arriving. NIST is developing stress tests that include short and ambiguous sequences, while future screening systems will increasingly have to deal with sequences modified or designed by AI rather than copied directly from known pathogens.
Policy is moving too. Federal rules now tie some US research funding to the use of synthesis providers that screen orders, while the broader federal oversight framework for high-risk life-science research has been rewritten again this year.
The technical difficulty is keeping false alarms low while still catching sequences that genuinely deserve review. Decentralized synthesis makes that harder because screening a limited group of large DNA companies is much easier than ensuring that thousands of independent machines apply strong screening consistently.
This chart, featured in our synthetic biology market deck, compares the main business model options for synthetic biology platforms
So what are the biggest unsolved problems in synthetic biology today?
The biggest unsolved problem in synthetic biology today is predictable engineering: we still cannot reliably turn a desired biological behavior into a design that keeps working across cells, environments, time and industrial scale.
After looking across genome writing, minimal cells, industrial fermentation, AI, medicine and ecological engineering, that problem sits above the rest.
Predictable scale-up comes next. Synthetic biology can produce spectacular results in controlled experiments and still lose enough performance during manufacturing to ruin the economics. The latest industrial research keeps finding the same issue even as strain engineering improves.
Our incomplete map from genotype to phenotype sits underneath both problems. The minimal-cell work makes that unusually hard to ignore. Researchers have built and modeled one of the simplest living cells ever made, yet biology at that scale still contains unknown functions and unexpected interactions.
Genome writing exposes the same gap in a different way. We can now build synthetic chromosomes and even assemble megabase-scale mammalian DNA. Debugging those designs still teaches us biological rules after construction that we would ideally have known beforehand.
Evolution adds another layer. An engineered cell can change after we finish building it, especially when our desired function makes the organism less fit. Long-term biological engineering therefore has to anticipate adaptation rather than treat the original sequence as a permanent specification.
AI will probably reduce the practical cost of all of these problems faster than any other current technology. The OpenAI–Ginkgo experiment already shows how much 36,000 automated tests can accomplish. Autonomous labs can compensate for imperfect prediction by searching biological space at a speed humans could never match manually.
Medicine is beginning to show what happens when the tools get good enough. Logic-gated cell therapies have reached real patients, and gene and cell therapies keep expanding clinically. Delivery and reliability still decide which synthetic programs can actually become treatments.
Environmental synthetic biology remains further from routine deployment. Gene drives are approaching their first true field tests while resistance, ecosystem behavior and post-release monitoring remain open questions.
Biosecurity closes the loop. As biological design, AI and DNA synthesis become easier, the systems that govern access and detect concerning sequences have to improve at the same time.
Most of the list collapses into one recurring problem. Synthetic biology can increasingly build biology, but it still struggles to know exactly how that biology will behave before the building starts.
| Rank | Biggest unsolved problem | Where synthetic biology stands now |
|---|---|---|
| 1 | Predicting biological behavior from a design | Still heavily dependent on experimental iteration |
| 2 | Predictable industrial scale-up | Lab performance often changes inside large bioreactors |
| 3 | Connecting genotype to phenotype | Even minimal cells still contain poorly understood biology |
| 4 | Long-term evolutionary stability | Engineered functions can lose out to faster-growing mutants |
| 5 | Whole-genome design and debugging | Genome-scale construction is real; reliable first-pass design is still rare |
| 6 | Robustness across environments and organisms | Designs remain strongly dependent on host and conditions |
| 7 | Delivery inside the human body | Several tissues are accessible; general cell-specific delivery remains elusive |
| 8 | Reliable programmable medicines | Complex gene circuits have entered trials but remain early |
| 9 | Low-cost commodity biomanufacturing | Economics demand exceptional titer, rate, yield and recovery |
| 10 | AI models that predict rather than mainly search | Autonomous experimentation is advancing faster than whole-cell prediction |
| 11 | Safe ecological engineering | Gene drives remain pre-field-trial technologies |
| 12 | Cheap DNA synthesis with strong biosecurity | Screening is improving quickly while synthesis capability keeps expanding |
If you want more recent data on this point, please see our latest synthetic biology market report.
OUR METHODOLOGY
This analysis asks which problems still most seriously limit synthetic biology today. We separated the field into predictive design, minimal-cell biology, genome writing, evolutionary stability, environmental robustness, industrial scale-up, autonomous experimentation, programmable medicine, ecological engineering and DNA-synthesis biosecurity, then compared where the same bottlenecks appeared across several of those areas.
We prioritized recent primary research, official scientific and regulatory material, first-hand technical results and high-level peer-reviewed reviews. Individual demonstrations were treated as evidence of what is technically possible, while the final ranking gives more weight to problems that continue to appear across organisms, applications and scales.
Predictable engineering emerged as the main problem because it appears repeatedly in otherwise very different parts of the field. NIST’s Engineering Biology program and the U.S. National Security Commission on Emerging Biotechnology both identify predictive design and predictable scale-up as current priorities, while minimal-cell and synthetic-genome research show the same gap experimentally.
For fundamental biological understanding, key evidence came from the J. Craig Venter Institute’s JCVI-syn3.0 work, NIST’s work on the genetic requirements for minimal-cell division, and the Cell whole-cell model of JCVI-syn3A. Together these sources show how much biology can remain difficult to predict even inside an unusually small and well-characterized genome.
For genome writing and industrial engineering, we relied heavily on the Sc2.0 synthetic chromosome XVI study, the Nature Biotechnology analysis of lessons from Sc2.0, the INSTALL genome-writing study, the Nature Communications review of industrial-scale biomanufacturing, and Lawrence Berkeley National Laboratory’s analysis of titer, rate, yield and scale economics.
For newer applications, key sources included OpenAI and Ginkgo Bioworks on autonomous optimization of cell-free protein synthesis, Senti Biosciences on SENTI-202, FDA material on cell and gene therapy development, the PLOS Biology study of resistance-resilient gene drives, WHO material on preparations for gene-drive field testing, and NIST’s synthetic-nucleic-acid screening benchmarks.
The final ranking is an editorial synthesis rather than a mechanical score. We gave the highest positions to problems that constrain several downstream capabilities at once and remain clearly unresolved despite recent technical progress. That is why prediction, scale-up and genotype-to-phenotype understanding rank above narrower problems that may be decisive in one application but do not affect the field as broadly.
This chart, featured in our synthetic biology market deck, illustrates the share of revenue generated by each customer segment in the synthetic biology market
Related blog posts
- The most recent funding news in synthetic biology
- Which synthetic biology startup is growing the fastest?
- Synthetic Biology: who are the top startups?
- The main fundraising trends in synthetic biology
- How strong is fundraising in the synthetic biology market right now?
