Twist Bioscience Deep Dive: Silicon Throughput Is Earning the AI Wet Lab, but Profitability Still Has to Catch the Narrative
How the Company Makes Money
Twist Bioscience sells physical DNA and the products and services built on it. The economic engine is a proprietary silicon-chip synthesis platform that runs phosphoramidite chemistry at miniaturized scale, writing millions of distinct oligonucleotides in parallel rather than on conventional column synthesizers. That DNA is sold directly, assembled into genes and libraries, used as probes in next-generation sequencing workflows, or consumed internally to express and characterize proteins, mainly antibodies. Revenue is overwhelmingly product and service, not royalty or instrument. Fiscal 2025, ended September 30, 2025, produced $376.6 million of revenue, up 20% from $313.0 million in fiscal 2024 and from $245.1 million in fiscal 2023. The five-year compound growth rate from fiscal 2020 to fiscal 2025 was about 33%.
Beginning in fiscal 2026 the company collapsed its old SynBio and biopharma lines into DNA Synthesis and Protein Solutions, and renamed NGS tools as NGS Applications. On the old product map, fiscal 2025 was NGS tools at $208.1 million, or 55% of revenue, synthetic genes at $113.6 million, antibody discovery at $23.5 million, oligo pools at $20.2 million, and DNA and biopharma libraries at $11.2 million. The new presentation makes the operating split clearer. In the quarter ended June 30, 2026, NGS Applications were $61.8 million, up 12% year over year, and DNA Synthesis and Protein Solutions were $56.6 million, up 39%. For the nine months, those lines were $171.9 million and $160.9 million. NGS is still the larger book; the faster growth, and the investment debate, now sits in synthesis and proteins.
Gross margin has been the second proof point after growth. It rose from 36.6% in fiscal 2023 to 42.6% in fiscal 2024 and 50.7% in fiscal 2025, then to 52.8% in the June 2026 quarter, with management stating that roughly 70% of incremental revenue in that quarter fell to gross profit. Full-year fiscal 2026 guidance, raised in August to $456 million to $457 million, implies about 21% growth, with gross margin expected above 52% and adjusted EBITDA breakeven targeted for the September quarter. The company is still GAAP-unprofitable. Fiscal 2025 net loss was $77.7 million. The June quarter net loss was $35.1 million. Cash, cash equivalents and short-term investments were about $167 million at June 30, 2026, against a shrinking but still negative adjusted EBITDA of $11.3 million in that quarter. The model is volume times a falling unit cost, not a software margin, and the operating leverage is only now approaching the line.
Customers, Competitors and the Supply Chain
The customer base is broad and sticky, which matters more than any single logo. Fiscal 2025 served more than 3,800 customers, with 99% of revenue from repeat buyers, and no customer above 10% of sales. By end market that year, healthcare contributed $215.1 million, chemicals and materials $93.2 million, academic research $65.9 million, and food and agriculture $2.4 million. In the June 2026 quarter the industry cut was therapeutics at $40.4 million, up 49%, diagnostics at $43.8 million, up 15%, academic and government at $15.5 million, up 32%, with industry and applied and global supply partners making up the balance. The therapeutics acceleration is the new fact. Diagnostics remains the larger and steadier NGS-driven pool.
Concentration exists inside the NGS book even if it does not exist at the company level. Top-ten NGS customers have recently accounted for roughly 39% of NGS Applications revenue, down from the low-to-mid 40s a year earlier, and the identities rotate. These are clinical assay developers whose commercial volumes, not research budgets, now drive a large share of NGS growth. Antibody discovery partnerships have included Bayer, Boehringer Ingelheim, Takeda, Astellas and Ono. As of September 30, 2025, Twist had 442 revenue-generating partnerships, 1,182 completed programs and 84 active programs, 82 of which carried milestones or royalties. Milestone economics are still a small slice of revenue; the partnerships matter more as a funnel into gene, library and characterization orders.
Competition is different in each book, and Twist is not the leader in either on a pure revenue basis. Merchant DNA synthesis is fragmented. Industry surveys consistently put GenScript, Thermo Fisher’s GeneArt business, Danaher’s Integrated DNA Technologies, Twist and a fifth player such as BBI or Eurofins in a top group that together holds roughly half the market. IDT dominates high-volume custom oligos and CRISPR reagents through Danaher’s procurement channels. Thermo wins on catalog breadth and one-stop distribution. GenScript competes aggressively on genes, especially for price-sensitive and China-linked demand, and has added GMP capability. Eurofins and Azenta’s GENEWIZ compete on turnaround in regional service models. In NGS sample preparation the named rivals are Illumina, Agilent, Roche, IDT, Thermo Fisher, New England Biolabs and Watchmaker Genomics. Agilent’s SureSelect franchise is the historical incumbent in target enrichment. Twist’s edge there is probe uniformity and the ability to print highly customized panels at a cost and lead time array chemistry struggles to match. In antibody discovery the relevant specialist comparator is Adimab, which sells access to a proprietary yeast platform and keeps more of the downstream economics; Twist is a high-throughput synthesis-and-expression shop, not a partnered discovery company in Adimab’s mold.
Supplier risk is low relative to most tools companies. The differentiated asset is the writer and the process around a silicon chip that Twist designs and fabricates, not a single purchased component. Reagents are phosphoramidite chemistry that has been commercial since the 1990s. The more interesting supply fact runs the other direction: Twist has historically sold synthetic DNA into competing synthesis companies, which is only rational if its unit cost sits below theirs. Capacity is split between South San Francisco, where chip fabrication, platform development and much of NGS manufacturing sit, and Wilsonville, Oregon, which automates the downstream synthetic-biology workflow for genes and therapeutic materials.
Where the Advantage Actually Sits
The chemistry is not proprietary. Emily Leproust has said as much: the reactions have been known for decades, and many Asian providers can write DNA. The moat is the machine. A chip the size of a conventional plate carries on the order of a million synthesis sites, organized so that a production chip can yield on the order of 10,000 genes or about a million oligos. Installed writers support a stated capacity of roughly 32 million oligos a day, and management has indicated the fleet runs well below theoretical capacity. Miniaturization cuts reagent use and waste, which is the cost advantage, and parallelism is the customization advantage. A customer ordering 100 antibodies has several suppliers. A customer ordering 1,000 or 10,000 sequence variants, each needing DNA, expression and a comparable data package, has very few.
Quality and speed have caught up with cost, which was the original wedge. Management has described production error rates around one in 2,500 to one in 3,000 base pairs, versus roughly one error per 100 bases on traditional column synthesizers at practical settings. Express clonal genes ship in four to seven business days, against a historical 30-plus-day cycle when the gene product launched and a still-common 10-to-15-day standard elsewhere. On an internal oligo dataset running from early 2023 through early fiscal 2026, manufacturing cost fell about 60%, chemical waste fell about 70%, turnaround on that workflow fell from 26 hours to 7 hours, and capacity rose about fourfold. Those are process numbers, not product prices, but they explain how gross margin could expand eight points in fiscal 2025 and another two points into the mid-50s while volume kept rising.
The advantage compounds when Twist sells the next step, not just the DNA. Genes feed antibody expression, purification and characterization. Oligos feed target-enrichment panels that sit inside customers’ commercial diagnostic kits. The same infrastructure therefore supports three revenue types with different durability: one-off research orders, recurring commercial-assay pull-through, and larger design-build-test campaigns from AI drug-discovery groups. Repeat revenue at 99% is the empirical result. The limitation is equally clear. This is not a network effect and not a switching-cost monopoly. A pharma account can dual-source standard genes from IDT or GenScript tomorrow. What is hard to dual-source is a campaign of thousands of variants with consistent quality, a 12-day complex-gene turnaround, and a characterization dataset that an external model can train on.
Industry Dynamics: Demand Pull and Competitive Pressure
Two demand curves are pulling the business, and they should not be conflated. Diagnostics, especially liquid biopsy, therapy selection and minimal residual disease, is a volume business tied to test adoption, reimbursement and the sequencing installed base. Twist does not own the assay or the sequencer. It sells library prep and custom probes into assays that other companies commercialize. That makes NGS Applications a high-quality, somewhat GDP-plus tools franchise with periodic air pockets when a large customer transitions from development to commercial scale, exactly the pattern management flagged heading into fiscal 2026. NGS growth slowed from 23% in fiscal 2025 to 12% in the first three quarters’ run-rate of fiscal 2026, even as diagnostics industry revenue in the June quarter still grew 15%. MRD is growing faster than the NGS book but remains a small share of it. Initial fiscal 2026 planning assumed only one to two points of NGS growth from MRD, with the real ramp pushed into late 2026 and 2027. The opportunity is large if personalized panels become standard of care. The threat is that panel content standardizes, or that Illumina, Agilent or Roche bundle enrichment tightly enough to cap Twist’s price.
Therapeutics is the demand curve the equity market is underwriting. AI models now propose far more antibody, enzyme and nucleic-acid sequences than wet labs can make and test. Twist’s own framing is that by 2030 the serviceable market in DNA synthesis and protein solutions exceeds $7 billion inside a total addressable market near $18 billion, with NGS adding a separate multi-billion-dollar pool, and a combined serviceable market on the order of $13 billion. Those figures are company estimates and should be treated as ceilings, not forecasts. What is observable is order behavior. AI-linked orders were more than $25 million in fiscal 2025 against essentially nothing the year before. Management has said it expects triple-digit percentage order growth in that category in both fiscal 2026 and fiscal 2027, and has indicated AI-related revenue could roughly double from 2026 into 2027, toward $100 million. Orders are not revenue, and “AI-enabled” is a management definition. The supporting evidence is physical: gene shipments in the June quarter were 369,000, up 56% year over year, after 32% growth in each of the prior two quarters. That is harder to fake than a pipeline slide.
The threats are price, mix and substitution at the margin, not a collapse in demand for synthetic DNA. Standard genes are contestable. Chinese and European service providers will keep pressure on list price for straightforward constructs, and Twist’s historical $0.09 per base pair positioning was a share-gain tool that cannot be the long-term margin story. Academic and government budgets remain cyclical; the 32% rebound in that bucket in the June quarter followed a softer period and should not be annualized. Biosecurity screening of synthetic DNA orders is a rising compliance cost and a potential friction on the fastest customers, though it also raises the barrier for informal suppliers. The structural risk to the bull case is that AI drug discovery produces more designs but fewer funded wet-lab campaigns if biotech financing tightens, or that model builders vertically integrate routine synthesis once volumes justify it. Neither has shown up in the order book through the summer of 2026. Both belong in any multi-year underwriting.
Products That Can Move the P&L
Three initiatives have a credible path to material revenue, and one does not. Complex Genes, launched from early access in May 2026, extends clonal synthesis to 7,000 base pairs and to sequences with homopolymers, repeats and extreme GC content that chemical platforms historically rejected. Twist now expects to accept about 99.5% of clonal gene orders and 99.9% of DNA product requests, up from roughly 96% of clonal genes three years ago and about 97% immediately before the launch. Early access produced more than 1,800 constructs across more than 100 orders, most delivered inside a 12-day window, on the same automated line as standard genes. Management has described the near-term revenue impact as a low-single-digit share gain on a base of roughly a million genes shipped per year. That understates the strategic point. AI-designed proteins and nucleic-acid therapeutic elements are disproportionately complex. If Twist can take those orders without a separate factory, it keeps the high-value tail that enzymatic specialists are trying to own, at a cycle time those specialists have not matched.
MRD Express is the diagnostics analogue. Personalized residual-disease tests need a different probe panel for each patient, which is an awkward fit for array printing and a natural fit for a million-site chip. The product in development targets roughly one-day turnaround and a dedicated capacity on the order of 1.15 million personalized panels and about 2.3 billion probes a year. Twist already supplies library prep and enrichment for both tumor-informed and tumor-naive MRD customers. The incremental profit pool is the conversion of that relationship from research panels into high-velocity clinical logistics. Timing risk is real: assay developers, not Twist, control regulatory clearance and reimbursement, and fiscal 2026 was explicitly not underwritten on a large MRD contribution.
The September 2026 Lilly TuneLab agreement is a distribution and data channel, not a disclosed revenue contract. Twist is a provider of antibody characterization data for TuneLab, including AbLab, Lilly’s developability model, using preferred protocols and preferred pricing for participating biotechs, with data flowing back into federated training. TuneLab’s models were trained on preclinical datasets Lilly has described as costing more than $1 billion to generate. Combined with Anthropic’s use of Twist wet-lab services to check AI-designed proteins, the pattern is that model owners need a scaled, consistent experimental layer and are willing to route customers to one. The February 2026 license of Invenra’s B-Body bispecific platform, paired with a 6.24% fully diluted equity stake, extends the protein menu toward formats pharma actually develops. mRNA products are on the roadmap and not yet a revenue line. DNA data storage remains a research collaboration, including work with public programs, and is not a planning assumption for the 2031 revenue target of more than doubling organic sales with gross margin above 60%.
Credible New Entrants
Two enzymatic challengers are real companies with customers, and neither yet threatens the core P&L. Ansa Biotechnologies commercialized enzymatic synthesis using terminal deoxynucleotidyl transferase and, in October 2025, launched a 50-kilobase sequence-perfect clonal DNA product, the longest commercially available, priced from $0.28 per base pair with turnaround under 25 business days and a reported fulfillment rate above 97% on long, complex constructs. It has raised growth capital and is moving U.S. manufacturing toward continuous operation. That product wins where Twist’s menu stops. Leproust’s response has been precise rather than dismissive: Twist’s standard clonal offering up to 5 kilobases, now extended toward 7 kilobases on complex genes, covers the vast majority of orders, and on that range chemical synthesis is still faster and cheaper. Ansa is a threat to the long-construct niche and a potential future threat if enzymatic cost curves cross chemical cost curves below 10 kilobases. It is not, in 2026, a substitute for a customer who needs 10,000 variants in a week.
DNA Script is the other credible enzymatic company, and its strategy is orthogonal. The SYNTAX platform is a benchtop printer that makes up to 384 oligonucleotides in a lab, with distribution now extending into India and other growth markets, plus an ambition in GMP template DNA. That attacks lead time and supply-chain control for small oligo batches. It does not attack Twist’s parallelism, NGS probe scale, or antibody characterization throughput. Molecular Assemblies, the other enzymatic name investors used to cite, shut down in 2024 and its intellectual property moved to Maravai, which removes one competitor and does not create a scaled one. The disruptive risk to monitor is Ansa, or a well-funded enzymatic platform, closing the cost and speed gap on the 1-to-7-kilobase gene that is Twist’s volume product. That gap has not closed.
Management’s Record
Leproust has run the company since co-founding it in 2013 and has chaired the board since 2018. The operating record from a standing start is unambiguous: revenue from $2.3 million in fiscal 2016 to $376.6 million in fiscal 2025, fourteen consecutive quarters of sequential growth through June 2026, and a gross-margin rebuild of more than 16 points in three years. Fiscal 2026 guidance has been raised three times, from an initial $425 million to $435 million in November 2025, to $435 million to $440 million in February, to $442 million to $447 million in May, and to $456 million to $457 million in August, with each reported quarter beating the revenue range then in force. That is a management team that has under-promised on the top line through the current cycle, not one that has had to walk numbers back.
The weaker part of the record is the long delay between scale and profit. Operating losses remained above $200 million as recently as fiscal 2024, and adjusted EBITDA was still a $46.9 million loss in fiscal 2025 despite the margin expansion. Breakeven has been a rolling target that is only now sitting inside a guided quarter. Patrick Finn, as president and COO, owns the factory and the product launches that made Express and Complex Genes manufacturable on one line. Adam Laponis, as CFO, has kept the cash trajectory compatible with that breakeven claim without an equity raise in the recent window; an at-the-market program of $200 million remained unused at June 30, 2026. Siyuan Chen, the chief technology officer and the company’s first R&D employee, is the continuity on the silicon process.
Capital allocation has been mostly internal: automation, a modest Invenra license and stake, and commercial hiring of about $10 million per quarter that management said would not jeopardize the fiscal 2026 adjusted EBITDA commitment. The governance blemish is personal selling into the re-rating. A Form 144 disclosed that Leproust sold about 552,000 shares for $82.2 million between August 3 and September 24, 2026. That does not change the operating math. It does mean the person with the most information reduced exposure while the equity narrative shifted from a tools turnaround to an AI infrastructure story. Investors should underwrite the factories and the order book, not the alignment implied by a rising share price.
The Scorecard
Twist is the scaled merchant of synthetic DNA that AI drug discovery and personalized diagnostics both need and cannot easily replicate. Silicon parallelism, a falling unit cost, Express and now Complex Genes, and an integrated path from sequence to characterized protein are a genuine advantage against IDT, Thermo, GenScript and Agilent in the jobs that require thousands of variants or patient-specific panels. Fiscal 2026 is on track for roughly $457 million of revenue, gross margin above 52%, and the first quarter of adjusted EBITDA breakeven, with therapeutics revenue growing near 50% and gene shipments up more than 50% in the latest quarter. The Lilly channel and the AI order commentary, including a path toward roughly $100 million of AI-related revenue in fiscal 2027, are consistent with the shipment data rather than a substitute for it.
The investment tension is that the easier work is done and the harder work is not. NGS, still half the company, has decelerated to low-teens growth and depends on a handful of diagnostic customers and a still-small MRD ramp. GAAP losses and cash consumption have not ended. Enzymatic platforms own the ultra-long niche and will press on complexity. Standard-gene pricing remains contestable. Management’s 2031 aim of more than doubling revenue at above 60% gross margin is achievable if AI campaigns and MRD logistics both scale, and it is a mid-teens growth plan rather than a hypergrowth plan once the higher fiscal 2026 base is the starting point. The platform deserves respect. The durability of triple-digit AI order growth, the conversion of that demand into gross-margin dollars above 55%, and a full year of positive cash earnings are the facts that still have to arrive.