Ginkgo Bioworks Wins Four University Autonomous Lab Orders and Undercuts Offshore CROs on Price, but Revenue Fell 48% and Visibility Remains Thin
Q2 2026 earnings call, August 5, 2026: new academic wins and ADME-One launch overshadow another steep revenue decline
Ginkgo Bioworks used its second-quarter call to move the debate away from the shrinking services base and toward its autonomous lab strategy. The company announced the day before the call that it was selected to build autonomous labs for MIT, Caltech, the University of Maryland and Northwestern. Caltech, Maryland and Northwestern are funded through an NSF program, and MIT through a separate grant. The wins add commercial evidence to a thesis that has so far rested largely on one national lab contract. Management gave no dollar values, no backlog and no revenue guidance, so the size of the opportunity cannot yet be judged.
University wins broaden the autonomous lab customer base
The four awards span distinct research areas. Caltech will focus on chemical structure data generation from natural-world compounds, Northwestern on protein engineering, Maryland on biomanufacturing and MIT on education. CEO Jason Kelly stressed that all four will run on the same underlying platform. Universities, he argued, share roughly 60% to 70% of their equipment, with only 30% to 40% specialized by field. His pitch is aggressive: "The proposal is there should be a giant Autonomous LabCorp in every biology department, and you could kind of close most of the labs down. It would feel a little more like buying time on a data center."
Kelly pointed to NIH spending of roughly $40 billion a year on research, plus NSF and DARPA budgets, as the pool of money in play. He expects "a good amount of FOMO" among institutions without these systems if the first batch performs. He also argued that universities could fund purchases from lab overhead budgets or donors rather than only from new grants. That is a plausible argument but an unproven one. Academic procurement cycles are slow, and the customers here are largely dependent on federal funding.
Kelly cited the White House Office of Science and Technology Policy's new science strategy, which includes a $400 million NSF program for a national network of cloud laboratories and explicit language on closed-loop autonomous experimentation. Ginkgo also pointed to its earlier work at Pacific Northwest National Laboratory, where the first 13 RACs were cut over in December, with a planned expansion to 97 under the Genesis Mission.
ADME-One is a price attack on WuXi and Western CROs
The most concrete new commercial product is ADME-One, launched about six weeks ago. It is a panel of five Tier 1 absorption, distribution, metabolism and excretion assays for small molecules. Ginkgo prices it at $199 through its Datapoints unit. Kelly said Western CROs charge $2,000 to $5,000 for the panel and Chinese vendors $1,000 to $2,500. Partnerships with Inductive Bio for pharmacokinetic projection and Tangible Scientific for compound management bundle the offering for "close to a tenth of price."
Management said it has validated the assays against external vendors, using Spearman rank correlations across kinetic solubility, permeability, microsomal stability, P450 inhibition and plasma protein binding. It also published data on the 320-compound LOPAC library. Kelly said an outside firm, Inflexa AI, reviewed that data and found the platform technically clean. His summary was blunt: "We think we stand up very well to WuXi in terms of technical capability and throughput and we kick their butt on price."
Early traction is notable but unquantified. Kelly cited about 16 customers in the first six weeks, some of them new logos. He highlighted why the category matters: ADME work is ordered continuously as pharma teams develop molecules, unlike AI-driven data generation, which tends to arrive in campaigns. "I'm also pretty excited to go after the traditional CRO because it's just a reliable source of demand. But we've got to prove ourselves, we're new in that area." At $199 a panel, the revenue contribution is likely small near term, and Ginkgo did not discuss unit margins. Plate-based chemistry and anaerobic chambers are the next planned additions, which would extend the offering into chemical synthesis, a larger share of China-based CRO revenue.
Nebula doubled in three weeks
Ginkgo's Boston lab, Nebula, now has 105 RACs after the company added 50 over roughly three weeks. Kelly said the RACs were prebuilt in advanced manufacturing, then installed, connected and debugged before running experiments. He called that timeline unheard of in conventional work cell automation. On a typical day, about 30 unique protocols submitted by scientists, rather than automation engineers, run across the system, and with copies across devices the count exceeds 100. Kelly acknowledged the system is not bug-free: "It's not like we don't have bugs, we don't have issues to work through, we do." The scheduler and robustness are the main engineering priorities.
Kelly framed the economics with a transportation analogy. Work cells are subways, highly automated but inflexible. Manual benches are cars, flexible but human-dependent. Ginkgo is trying to build the Waymo, with automation and flexibility combined. He said benches account for more than 95% of the $60 billion to $80 billion pharma spends annually on research. He also said an autonomous lab fits equipment into about one-third of the space and runs 168 hours a week versus roughly 40 for a manual lab. The company is also offering a try-before-you-buy option, running prospective buyers' protocols on Nebula before they commit to a system.
Financials: revenue down sharply, burn guidance holds
The core business is still shrinking. Second-quarter revenue from continuing operations, which now excludes the biosecurity unit sold on April 3, was $20 million, down 48% year over year. First-half revenue of $40 million was down 49%, or about 42% excluding $7.5 million of noncash BiomEdit termination revenue in the prior-year period. Adjusted EBITDA was negative $36 million versus negative $25 million a year earlier. Net loss from continuing operations was $57 million versus $53 million. R&D slipped 4% to $30 million and G&A fell 26% to $12 million, reflecting restructuring completed at the end of 2025.
Cash burn was $45 million in the quarter versus $38 million a year earlier, and $93 million for the first half, down 3%. The first half included a $14 million Google Cloud payment. The reset of that commitment cut minimum obligations by more than $100 million and extended the term from three years to six. Adjusted EBITDA also carries $14 million a quarter of excess lease costs, up from $12 million, which management said could be mitigated through subleasing.
The company reaffirmed 2026 cash burn guidance of $125 million to $150 million. That implies second-half burn of roughly $32 million to $57 million, a sharp step down from the first half, and Kelly said burn should improve further in the second half. Ginkgo ended the quarter with $302 million in cash and equivalents plus $87 million in restricted cash. It also raised $17 million through its at-the-market program in the quarter, a reminder that dilution remains a funding lever.
Revenue visibility remains the key gap
Analysts, including TD's Brendan, pressed for revenue detail on autonomous labs and Datapoints, and management gave little. Ginkgo does not provide revenue guidance and does not break out autonomous lab or Datapoints revenue. CFO Steve Coen said hardware revenue is recognized at delivery and installation, even if cash arrives earlier. The large government project is still in what he called the "planning coordination phase," with only small revenue from a preliminary contract. Services and software licensing follow after deployment. Datapoints revenue is recognized over time on projects of three to nine months, which Coen said are smaller than historical solutions deals.
The implication is that lumpy equipment revenue from the PNNL expansion and the university labs will arrive only after installs, while the legacy services base keeps declining. Management also declined to disclose repeat order rates, saying only that deals typically begin with a proof of concept and expand if customers are satisfied. Investors are being asked to underwrite a pivot with strong anecdotal momentum but little disclosed financial evidence.