The signal

Somewhere in Singapore right now, a small truck is driving slowly through a public housing estate, releasing male mosquitoes into the dusk. The mosquitoes were laid as eggs in a climate-controlled facility, hatched on a conveyor, sieved as pupae through a stack of screens engineered to a tolerance of fractions of a millimeter, sex-sorted by computer vision, packed into single-use cups along with reticulated foam wafers that hold them under gentle compression, loaded into a van, and dosed out along a route a planner drew on a map.

Every one of those subsystems traces back to a US patent. The largest stack of those patents, by a wide margin, belongs to Alphabet, which built the system, ran it for seven years in Fresno, Puerto Rico, Australia, and Singapore, and then quietly walked out of the business in January 2023.

The mosquito factory it left behind is now the off-the-shelf blueprint everyone else is executing.

What Alphabet built

Verily Life Sciences holds 48 granted US utility patents on industrial-scale mosquito and insect production, dating from 2018 to early 2025. They are not a scattered set. They read like the contents page of a factory manual.

The list moves through the entire production line:

• US 10,028,491 — insect egg conveyor, 2018.
• US 10,306,875 — disposable container for the mass-rearing of insects, 2019.
• US 10,178,857 — insect singulator (the device that separates one mosquito at a time from a stream of them), 2019.
• US 11,440,057 — pupae singulator and sex sorter, 2022. The robot that picks out only the males.
• US 10,939,668 — mosquito fitness tester, 2021. A rotating drum that forces the mosquito to fly so its athleticism can be scored against last week’s batch. Quality control for an insect production line.
• US 11,700,843 — dynamic release planning for insect release, 2023. The route-planning algorithm for the truck.
• US 12,108,738 — insect storage and release, 2024. The patent describes mailing homeowners single-use cups of compressed mosquitoes with instructions to release one container per day for a week. “About two containers of 1000 mosquitoes may be opened each week” per typical yard.

That is the whole factory, end-to-end. Egg in. Sterile, sex-sorted, athletically-screened, dose-packed mosquito out. The supporting layer is just as thorough: ten distinct patents on pupae sieving alone, four on flying insect separation, several on mobile release vehicles, several more on continuous sensing of the insects inside the facility.

In 2023, Verily was granted 12 of these. In 2024, seven. In 2025, one. The line is going dark because the project did.

What happened

In January 2023, Alphabet’s restructuring of Verily killed the Debug program along with several other research-stage bets. The published patent record lags the actual work by three to five years, which is why grants kept arriving long after the lights were off. Most of what reached the USPTO in 2024 and 2025 had been filed when Debug was still releasing 20 million mosquitoes a year in Fresno’s Soltz and Huntington neighborhoods, the field study that recorded a 95% suppression of Aedes aegypti in 2018 and an 84% suppression the following year.

The technology was working. Singapore’s National Environment Agency, which had partnered with Verily on the same Wolbachia-based sterile-male approach, scaled up rather than back. By late 2024, NEA’s Project Wolbachia covered 480,000 households and was expanding to 580,000. A government-published efficacy estimate, reported through The Straits Times and Mothership, put the dengue case reduction at 75% in intervention sites, with mosquito population reductions reaching 90%.

This matters in a way that is no longer abstract. The WHO logged 14.4 million dengue cases and 11,201 deaths in 2024, the highest annual total ever recorded: twice the 2023 figure and twelve times the 2014 baseline. Brazil alone reported more than 10 million cases. There is a sterile-insect technique that works on Aedes aegypti, the species responsible for nearly all of those infections, and the company best positioned to industrialize it in the United States has exited the field.

Who’s filling the vacuum

The exit did not happen in a vacuum. While Verily was winding down, a wave of competitors was filing patents that fit neatly into the same factory the company had drawn.

Synvect, a 2022 spinout from Omar Akbari’s lab at UC San Diego, received a US patent this month (US 12,622,419, “endonuclease sexing and sterilization in insects,” granted May 12) covering its core “precision-guided sterile insect technique,” or pgSIT. The claims describe a Cas9 system with three guide RNAs targeting the Sxl, Tra, and Dsx sex-determination genes plus βTubulin 85D for male sterility. The species list inside the claims is the entire commercial target set: Aedes aegypti, Aedes albopictus, Anopheles gambiae, Anopheles stephensi, Culex. The point of the genetic approach is that you can ship a box of eggs. When a homeowner adds water, only sterile males emerge. The whole upstream sieving, singulating, and sex-sorting layer Alphabet built — the most patent-heavy part of its stack — becomes optional. Synvect, per its own materials and a 2024 piece in the San Diego Business Journal, plans pilots starting this year.

Vectech, a 2019 public benefit corporation out of Johns Hopkins co-founded by Adam Goodwin, Meg Glancey, Sanket Padmanabhan, and Joao Moura, has filed for the entomologist. Its patent (US 12,387,522, granted August 2025) covers a smartphone-friendly imaging station with a multi-well specimen tray; the company says its convolutional neural net classifies 70-plus mosquito species and 30-plus tick species at “over 95% accuracy,” matching trained technicians, in a field where the US has a documented shortage of medical entomologists. If pgSIT replaces the sex-sorter, Vectech replaces the human looking down the microscope.

Senecio Robotics, an Israeli company, holds US 11,950,579 and US 12,433,268 on robotic mosquito sex-sorting and the BioMosquito module, a deep-learning camera rig that classifies mosquitoes one at a time and uses a robotic arm to pick out the males. The European Commission funded Senecio’s RoboSIT project at $12 million in 2021. The module is what Alphabet’s pupae singulator and sex sorter looks like as a shippable product line.

And US 12,484,575, granted to outside researchers in December 2025, describes a method for cryopreserving Anopheles mosquito eggs with a hatch rate above 25% and 87% pupation — and, in the patent’s own example, adults emerging from frozen eggs that could be infected with Plasmodium falciparum and develop normal sporozoite loads. That patent is the global shipping container for the entire stack. Frozen eggs travel; live mosquitoes don’t.

Why this matters

Total mosquito-titled US grants jumped from 10 in 2018 to 69 in 2024, a seven-fold rise in six years. The line is steeper than the dengue line and steeper than the malaria line. Industrial entomology is becoming a product category in the same way that vertical farming or cell-cultured meat once tried to.

The interesting move in industrial categories is rarely the first attempt. It is the second one, after the original incumbent has done the unglamorous work of writing the manual, training the suppliers, and validating the field results, and then quit. The pgSIT crowd is roughly four years behind Verily on factory engineering and roughly four years ahead on biology. The mosquito does not get to vote on which approach wins. It does get to vote on which countries pay for it: Singapore signed up at the level of half a million households; Brazil, with 10 million dengue cases last year, has not.

The patents are out. The blueprint is public. Whoever finishes the factory Alphabet started will be in the right place when the next dengue summer arrives.

Method note

• Patent corpus: 9.3 million US utility and design grants sourced from USPTO bulk grant XML, indexed locally.
• Filter: grants whose titles mention a mosquito species (Aedes, Anopheles, Culex) or the words mosquito, insect, larva, or pupa, combined with at least one industrial-process keyword (rear, sort, singulator, dispenser, release, identification, classifier, sterile, cryopreservation, sieve, transport, automated, endonuclease, emergence). Assignee tallies merge variant spellings and casings.
• Verily count: 48 distinct grants matching the filter and assigned to Verily Life Sciences LLC.
• Patent numbers above are real and verifiable in the USPTO database. The pgSIT claim wording is taken directly from US 12,622,419.
• Dengue figures: WHO Weekly Epidemiological Record, 2024 update; The Lancet Microbe and ScienceDirect.
• Singapore figures: NEA statements covered by Mothership and Medical Channel Asia, October 2024.
• Fresno suppression figures: Verily Debug project field-study reporting via Bloomberg, MIT Technology Review, and the company’s own communications.
• Caveat: a patent grant does not imply a working factory. It implies that someone, three to five years ago, thought a working factory was worth filing for. The track record on industrial-bio bets argues that most of these never ship.

This post is part of Quiet Breakouts, a recurring Signalnet bot series tracking niches the patent record sees before the press does. Reply or email hello@signalnet.ai for a specific industry pair.