Can Python’s Creator Double Its Speed? Inside Guido’s New CPython Push

Python's creator claims he can increase performance 1‑5× without breaking any existing Python code—can he deliver?

Guido van Rossum, who created the popular programming language Python 30 years ago, has set a new ambition: to double Python's speed and address performance gaps compared with languages like C++.

He will tackle runtime speed within the CPython core, motivated by the rise of alternative Python runtimes such as JIT‑based Pyston.

At the recent PyCon conference, Guido posted a document on GitHub outlining his goal to make Python at least twice as fast in Python 3.11, with a pre‑alpha branch expected next year.

After being hired by Microsoft in November, Guido—who helped develop four million lines of Python at Dropbox—now serves as a Microsoft Distinguished Engineer, enjoying the freedom to choose projects.

Microsoft supports Python through free courses, Azure AI integrations, the VS Code Python extension, and the Windows 10 interpreter, and has been a sponsor of the Python Software Foundation since 2006, contributing $150,000 this year.

The company also funds a small Python team focused on performance improvements. Currently five core developers contribute to CPython at Microsoft: Brett Cannon, Steve Dower, Guido van Rossum, Eric Snow, and Barry Warsaw.

These developers are working on sub‑interpreters to enhance multithreading and overall performance. Security is also a priority, with Microsoft emphasizing protection of the PyPI package repository and supply‑chain integrity.

Microsoft’s Python lead, Dan Taylor, said the $150 k PSF sponsorship will target packaging work to improve PyPI and the ecosystem.

Guido promises that the changes will not break ABI or API compatibility, will not slow extreme cases, and will keep Python code maintainable.

While the team is unsure whether a 2× speedup will be achieved, they remain optimistic. Guido even envisions a five‑fold speed increase after Python 3.11, emphasizing the need for creativity.

The primary beneficiaries will be developers running CPU‑intensive pure‑Python code and web developers using Python. However, code already written in C extensions such as NumPy, TensorFlow, I/O bindings, multithreaded code, and inefficient algorithms may see limited gains.

Author: Luo Yi