Short answer: Pacific Northwest National Laboratory (PNNL) bought a modular, robot-run microbiology lab from Ginkgo Bioworks, then hired Ginkgo to build a much larger one under a four-year contract worth up to $47M. Both are funded by the US Department of Energy (DOE) Office of Science. It's a clear government demand signal: government is betting that modular robot labs enable research at unprecedented speed and scale. The business lesson is sharper: Ginkgo is capturing value by selling the automation, not the science.
What did PNNL buy from Ginkgo?
- First system: PNNL's Environmental Molecular Sciences Laboratory (EMSL) bought AMP2 from Ginkgo for $4.66M in July 2025 and commissioned it on Dec 4, 2025. It's 18 instruments in an oxygen-free chamber, described as "autonomous-capable," with automation engineers in the loop (EMSL; Ginkgo).
- Follow-on: a four-year contract worth up to $47M (a ceiling, not funded value) for a larger platform, M2PC, to open "by 2030" (Ginkgo; EMSL). Ginkgo CEO Jason Kelly calls it "7X bigger." Accounts of its size differ, so we're not giving a cart or instrument count.
- The Oct 29, 2026 post is an explainer video about this work (Ginkgo on X), not new hardware.
PNNL calls the facility the world's largest of its kind; that's its own assessment.
Faster and bigger, not "never possible before"
The accurate claim is speed and scale. Robot labs can run more experiments, more consistently, than people alone. How autonomous AMP2 is today, for example whether it already runs experiments an AI chooses, we haven't been able to verify. PNNL has also promised "publicly available," AI-ready data, which we haven't yet seen published.
The business caveat
Ginkgo's core business has struggled. It reported Q2 2026 revenue of about $20 million, down 48% year on year (Ginkgo Q2 2026 results, SEC, Aug 5, 2026). Selling the automation itself may fix value capture for this line of business, but it concentrates exposure to DOE budgets.
What it means for you
- Early-stage lab-automation founders: this is one template for a first government customer. Sell a working system, prove it, and earn a follow-on with a ceiling roughly ten times the first purchase.
- Investors: read it as land-and-expand, and as a pivot signal: the money may be in the reconfigurable hardware (our view). Watch DOE, National Science Foundation (NSF) and Genesis Mission budget lines as the demand signal, and weigh the concentration risk. Ginkgo describes the Genesis Mission as a White House effort to build autonomous laboratories, backed by $5 billion of investment (same SEC exhibit).
- CEOs and CFOs: if your product generates experimental data, ask whether you should sell the machine that makes it rather than the result.
- Engineers: real autonomy across many different instruments is still unsolved, and that's the interesting work.
The SoCal hook
Ginkgo says Caltech has "commissioned autonomous labs to be built by Ginkgo under the NSF's ~$400 million investment" in AI-enabled cloud labs (Ginkgo Q2 2026 results, SEC). That puts AI-programmable cloud labs in Southern California, too.
Open questions
- Does AMP2 run AI-chosen experiments today?
- When will PNNL publish the promised AI-ready datasets?
- Why didn't Ginkgo's captured experiment data turn into commercial value?
Your turn: what would your first government robot-lab sale look like? If you're building or buying self-driving labs, reply and tell us. Follow LA Robotics Network for more on science acceleration.

