Member-only story
How DeepSeek Will Change the Programming Industry
This can be GenAI’s Linux moment
When ChatGPT was released in 2022, it was a sensation. Within five days of launch, it reached the 1M user benchmark.
It was mostly hyped up in the non-tech areas: Writing novels, submitting college assignments, preparing sales pitches and marketing flyers. The software industry was among the last to be targeted.
Reason? GenAI has been a dollar-guzzling business. Non-tech customers were the easiest sales target.
Copilots became laughing stock. Only the most curious developers tried LLM-assisted coding on their personal devices. When it came to orgs, AI risks hindered the adoption. The more devs started using it, the more inaccuracies surfaced — some of them also driven by their fear of being replaced.
Tools like Cursor and a curious mix of competing chatbots (Claude and Gemini) have prepared programmers for the GenAI future. Yet, most of this interest is towards AI (let’s become ML engineer, big tech is hot on hiring!), not augmenting GenAI in their regular .NET/Java/Javascript/Cloudformation workflows.
Enter DeepSeek
DeepSeek’s V3 and R1 models were already published in Dec 2024. When its mobile app topped the App Store charts on 20th January, the world took notice.
When the world read its researchers’ claim of training it at $6M, it became a recipe for a true market disruption.
Newspapers were still starry-eyed looking at OpenAI’s $500B Stargate gala with Softbank, blessed by the new President. But on 27th January, markets screamed: $600B got wiped out, somewhat rebalancing NVIDIA’s unfathomable P/E ratio.
The world had to take notice of China’s power.
Some pictures are worth a million words (Inspiration: a Chinese Proverb)
0-No explanation needed, if you are following the meme world:
