How AI In-Ovo Sexing Could End Chick Culling

Newly hatched yellow chick, illustrating why AI in-ovo sexing matters for the egg industry

AI in-ovo sexing uses machine learning to tell whether a chicken egg will hatch a male or a female chick, days before it hatches. It is a clear example of what AI does best outside of chatbots: sorting real-world data into categories, quickly and at scale.

Key takeaways

  • Male chicks from laying breeds are not needed by the egg industry, so billions are culled each year after hatching.
  • In-ovo sexing finds the sex of the embryo inside the egg, so male eggs can be removed early.
  • AI classification models read MRI scans, light spectra or images of the egg to make the decision.
  • In fact, a Canadian-developed system can do this as early as day four of incubation.

AI is more than generating text and code

Generative tools get most of the attention. However, much of the practical value of AI comes from classification: looking at an input and assigning it to a category. For example, is this transaction fraud? Is this part defective? Will this egg hatch a male or a female? Although these systems rarely make headlines, they make thousands of decisions every hour.

What is in-ovo sexing?

"In ovo" is Latin for "in the egg." Unlike traditional chick sexing, which happens after hatching, in-ovo sexing identifies the sex of the embryo during incubation. Germany and France have already banned chick culling. As a result, hatcheries there needed alternatives, and the first commercial in-ovo sexing service launched in Germany in 2018.

How AI in-ovo sexing works

  1. Scan. First, eggs are scanned with MRI, hyperspectral imaging or spectroscopy.
  2. Label. Next, each scan is recorded as male or female once the outcome is known.
  3. Train. Then, a machine learning model learns the patterns that separate the two.
  4. Classify. Finally, new eggs are scanned on the production line and sorted automatically.

However, the model does not understand biology. Instead, it finds statistical patterns a human eye would miss and applies them consistently to every egg.

Who is doing it

CompanyMethodStatus
Orbem (Germany)Deep learning on 3D MRI scansCommercial; raised €55.5M Series B in January 2026
SANOVO, Canadian Egg Technologies and MatrixSpec (Mississauga)Hyperspectral imaging with machine learning, from day fourGlobal launch planned for late 2026 or early 2027
Omegga (EU funded)AI-powered absorption spectroscopyIn development, focused on lower cost

Germany, France, Italy, Switzerland and the Netherlands have moved fully to in-ovo sexing, and Norway will follow on July 1, 2027. Meanwhile, in the United States, producers such as NestFresh, Vital Farms, Kipster and Sauder's have adopted it. Nevertheless, commercial use of AI at large scale is still early, and cost and speed need to improve for global adoption.

The lesson for businesses

This approach also applies to fraud detection, medical image triage, crop disease detection and defect detection in manufacturing. The best AI projects are often not the most visible ones. Therefore, start with a specific, costly decision your organization makes often, and ask whether a model could make it faster, earlier or more consistently.

Have a classification problem in your business? Talk to Mantrax Software Solutions about building a practical AI solution for it.

Frequently asked questions

What is in-ovo sexing?

In-ovo sexing is a method to find the sex of a chicken embryo while it is still inside the egg, so male eggs can be removed before they hatch.

How does AI help with in-ovo sexing?

AI classification models are trained on labelled scans of eggs, such as MRI or hyperspectral images. The model learns patterns that separate male and female embryos and then classifies new eggs automatically.

How early can AI determine the sex of a chick?

A Canadian-developed hyperspectral imaging system can determine sex as early as day four of incubation. Other methods typically work later in incubation.

Is AI in-ovo sexing an example of generative AI?

No. Instead, it is an example of classification, a type of machine learning that assigns inputs to categories. It does not generate new content.

Sources

Photo by Afra Ramió on Unsplash

Recommended Posts