This is what you can consider an example of successful stories
Linux distros, aerospace companies and institutions such as NASA, or IoT (Internet of Things) and industrial machinery producers use Python for automation. YouTube, Spotify, Google's AIs, crawlers, and web pages are vastly run by Python. Companies like Facebook, Twitter, and Google released a nutritious amount of free software that they use (also by themselves) for anything from basic web tasks to data science (AI, analytics/stats, data mining, etc... that you already find even in basic packages of Anaconda packed with JupyterLab), and automation. I mean a lot of free but top, edge cutting software like e.g. these open source projects. Python is the main language of data scientific tools (perhaps, generally in science alongside R). Instagram used Django (Python), Haproxy, GraphQL, plus Redis; Twitter has deployed Haproxy, GraphQL, and uses also Python. The company behind GitHub (the biggest pages that host programming repositories) developed Haproxy, uses Python, and Redis. Websites of Pinterest, NASA, Mozilla, National Geographic, Disqus, Washington Post, The Onion, and many other companies (having an "in-house" employed tech guy), they use Django (Python). Reddit, PayPlug, Netfix, MIT University, Maligun, Cloudify, Zillow, Keen IO... use Flask (Python). Facebook uses Tornado powered by Python (Facebook doesn't limit itself only on PHP, but uses it too, slowly moving away) and with Yelp, Coursera, Google Cloud, Medium, Product Hunt, StackShare, Tumbir, and Artsy, all use GraphQL. Yahoo!, Airbnb, Uber, Stack Overflow (world's biggest community of programmers) use Redis and, partially, too Python.
Do I recommend others what I use by myself? Why?
Due to my studies in computer science (for a university degree) along with independent (e.g. this one Phytonic MOOC of multiple) MOOCs in data science, and own interests in automation, Python and networking have been a familiar way to go. Even, extensively, without any technical needs to utilize it all (but in an exploratory and learning/exercising manner that came before commercial and open source projects, this website itself and other back-end apps are run by multiple frameworks (e.g. Nickel) rather written in Rust and Python 3. Rust is new but the most favorized language between programmers which alternates C++ for higher speed and memory safety characteristics. To compare, see Nickel.rs (here, you find a full performance test if you scroll down), and Actix (check the benchmarks). Both of the frameworks are multiple times faster than PHP 7, faster than Go language frameworks and can handle more connections than Node.js. For example, I could mention that asynchronous Tornado web framework with Asyncio can be deemed as a Python alternative of Node and either Django, or Flask scale and more, are well suited with Multiprocess or Celery (apps written in PHP and many other languages can run only on a single CPU core/treat per an end-user). The main reasons for Python are scalability, multipurpose usability, available (free, open-source) libraries, and development speed/expenses (I worth my personal time and deadlines). In a planing manner for a future growth and adaptation to a changing environment (the environment will require a usage AI technologies, we'll minimally need to equal our tech stack with a competition), it doesn't really matter if you ask me, any random (good or bad) computer/data scientist, or well-priced employees of IBM, Google, Microsoft, or any other big company frequently working in automation, analytics, machine learning, advertisement, and other fields that are related or direct sub-fields of AI. Most often and reasonably, they consciously and helpfully advise you to stick with Python libraries and general connectivity to the language.