There’s a difference between making a bet, and making a move you can’t afford not to.
NASA knew which one they were making.
Behind the missions, the telescopes, and the frontier-breaking discoveries, there’s a quiet frontier most people never see: software infrastructure. Specifically, open source.
When NASA looked beneath the surface, they saw something no agency can ignore:
If the software backbone collapses, everything above it does too.
They asked the question every leader should be asking right now. What happens if the tools we rely on are no longer supported and maintained?
And they didn’t wait around to find out.
More than 90% of the software used in business, science, AI, and national security is open source.
Open source is the foundation of everything. And foundations crack without reinforcement.
So through its ROSES 2020 and 2024 programs, NASA funded critical efforts to stabilize and advance the scientific Python ecosystem—tools like NumPy, SciPy, pandas, and scikit-learn. These libraries are essential to groundbreaking work: climate modeling, anomaly detection, deep space simulations, and AI diagnostics across agencies.
NASA use cases for Tech Leadership
Space modelling
Climate Science
Anomaly detection
scikit-learn
SciPy
pandas
NumPy
Multiple core libraries were brought into better alignment: reducing integration issues, increasing trust in the stack, and enhancing workflows across data science teams.
Computational bottlenecks were addressed: accelerating key processes, introducing support for parallelization, and enabling broad scalability. That means faster turnaround, reduced infrastructure costs, and higher mission velocity.
Instead of relying on a shrinking bench of overworked volunteer maintainers, NASA’s investment expanded the pipeline of capable contributors for long-term project health.
No one should be renting mission-critical tools. Black box platforms limit visibility, customization, and adaptability.
Fixed length strings in NumPy and pandas
“International…” → “Internatio”
50-character allocations for 10-character strings
Mixed use of object dtype with inconsistent behaviour
Rigid string arrays leads to inefficiency in both storage and code maintainability
String-Dtype NumPy 2.0 and pandas
“Internationalization retained in full”
Dynamically sized allocations that fits real data
seamless handling via native StringDtype
Rigid string arrays leads to inefficiency in both storage and code maintainability
Let’s not dance around it.
NASA didn’t choose open source because it’s trendy.
They did it because it’s controllable, auditable, and built to scale with real needs.
01
controllable
02
auditable
03
built to scale with real needs.
The kind of needs that don’t wait.
The kind of needs your agency, department, or company probably feels right now.
What was built
GPU-enabled scipy.special function
Dask & CuPy integration in scikit-learn
Parallelized execution
Reduced latency in simulation modes
Optimized large scale array computation
Clear APIs, faster ranking
This isn’t a tech decision. This is the strategic move your operations need.
If NASA, the agency tasked with putting people on the Moon and robots on Mars, knows they must invest in open source—not because they want to—but because they must…
What are you doing?
How will you future-proof your operations?
