Panique sur les marchés : l'IA n'est pas la bulle, elle est l'épingle !

Panique sur les marchés : l'IA n'est pas la bulle, elle est l'épingle !

Panic in the markets: AI is not the bubble, it's the pin!

🎙 Renaud Dékode 👥 249K 📅 February 25, 2026 ⏱ 29 min 👁 25K 📄 news review 🧭 2026-09-07
Available in: English (current) Français

Keywords

IAbulleCOBOLDeepSeekOpenAI

Summary

In this news review, Renaud Dékode discusses several AI-related events from February 25, 2026. The main story is Anthropic’s publication about Claude Code’s ability to modernize COBOL applications, which caused IBM’s stock to drop by 13%, erasing $31 billion in market value. He argues that AI is not a bubble but a pin that bursts existing bubbles, such as IBM’s reliance on COBOL maintenance. He also covers DeepSeek’s rise, with allegations of training on sanctioned Nvidia Blackwell GPUs and industrial distillation of American models, raising concerns about export controls. Next, he highlights OpenAI’s gpt-realtime-1.5 and WebSocket support for the Responses API, improving real-time voice interactions and reducing latency. He then introduces Mercury 2, a diffusion-based LLM from Inception Labs, which generates text more efficiently than traditional autoregressive models. Finally, he mentions Google for Education’s initiative to train 6 million US teachers on AI, and the deployment of robot dogs for police support near Monterrey ahead of the 2026 World Cup. The video is informal and opinionated, with a focus on market impacts and technological shifts.

175 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides a valuable perspective on how AI is disrupting established industries, using the IBM COBOL example to illustrate the concept of AI as a ‘pin’ rather than a bubble. The argumentation is coherent and engaging, though it relies heavily on anecdotal evidence and personal interpretation. The discussion of DeepSeek’s potential impact and the diffusion-based Mercury 2 model adds depth, but the lack of concrete data or citations weakens the overall rigor. The presenter’s enthusiasm is clear, but the analysis would benefit from more balanced and evidence-based reasoning.

Scientific Rigor, Source Quality, Title Accuracy

The video does not cite specific sources beyond mentioning Anthropic’s papers and OpenAI’s documentation, and the description only provides links to the creator’s website and a paid subscription service. This limits the verifiability of the claims. The title accurately reflects the content, focusing on the market panic and the role of AI. The informal tone and lack of citations reduce the scientific rigor, but the video serves as a commentary rather than a formal analysis. No comments were provided, so public reception cannot be assessed.

189 words

Title / Content Match

The title accurately reflects the video's central thesis that AI is deflating existing market bubbles rather than being a bubble itself, as illustrated by IBM's stock drop.

Quality & Reliability

6/10

The video is a personal commentary on recent AI news, mixing factual reports with subjective analysis and informal language. It lacks citations to primary sources, and some claims (e.g., IBM stock drop, DeepSeek training on Blackwell) are presented without verification. However, the core events are plausible and align with known industry trends.

Chapters

Cited Sources

  • Renaud Dékode website — Mentioned as a place for discussion and additional content.
  • Klub subscription — Promoted for exclusive content and community access.

Concurring Sources

  • Anthropic's Claude Code announcement — The video references Anthropic's paper on Claude Code, which is likely the source of the COBOL claim.

Contribution & Novelties

The video offers a unique angle on AI market dynamics, framing AI as a disruptive force that pops existing bubbles rather than being one itself. It highlights specific examples like IBM’s COBOL modernization and the rise of diffusion-based LLMs, which are not widely covered in mainstream media. The discussion of DeepSeek’s potential training on sanctioned hardware adds a geopolitical dimension.

Pour aller plus loin :

  • COBOL — Background on the programming language central to the IBM story.
  • Diffusion models — Explains the technology behind Mercury 2.
  • Export controls on AI chips — Context for the DeepSeek GPU allegations.

98 words

Radar Profile

The radar profile shows moderate scores across all dimensions, with quantity of information being the highest. This indicates a video that provides a good amount of content but lacks depth in technical detail and reliability. The balance suggests a general audience rather than specialists.

Reliability 4/10