J.P. Morgan Warns of "Red Flags" in the AI Market — Parallels to the Dot-Com Bubble Pattern

On June 27, J.P. Morgan released an analysis containing multiple risk warnings about the AI market. According to the firm, 42 AI-related companies in the S&P 500 now account for 65–80% of the index's total earnings. The semiconductor sector's rally is exhibiting technical patterns observed during the dot-com bubble, and the market influence of leveraged chip ETFs has grown fivefold since early 2024.

J.P. Morgan points out that concentration risk is rising across multiple layers of market infrastructure and the broader economy, and warns of investor "exuberance." That said, the report does not deny the value of AI technology itself; rather, it questions the sustainability of the market structure.

Apple and Microsoft Raise Prices in Quick Succession Amid Memory Chip Shortage

On June 26, Apple and Microsoft announced price increases on major products within just five hours of each other. The affected products include the iPad, Mac, and Xbox consoles. Both companies cite an unprecedented memory chip shortage driven by the AI boom.

While high-profile efforts to expand supply are underway, Bloomberg reports that the shortage and its impact on consumer prices "won't be over anytime soon." Surging demand from AI data centers has started to spill over into the cost of consumer electronics.

Wall Street Bets Billions on AI's Power Problem

Facing a surge in electricity demand driven by the AI boom, Wall Street is accelerating investment in power infrastructure companies. In 2026 alone, at least 10 power infrastructure and cleantech companies have gone public through IPOs, raising more than $11.6 billion — a record pace for the sector.

Fervo Energy, a geothermal power company, jumped 35% on its first day of trading, a sign of just how high market expectations run. That said, some note that certain technologies are still not fully developed, so a speculative element is hard to deny.

DeepSeek Publishes New Paper "DSpark" on Accelerating Inference

DeepSeek has published a paper, "DSpark," on speculative decoding for speeding up LLM inference. It has drawn a huge response from the technical community, racking up 558 points and 216 comments on Hacker News.

Speculative decoding is a technique that improves inference speed while preserving accuracy: a small model rapidly generates candidate tokens, which a large model then verifies. DeepSeek has a track record in efficient inference technology, and attention is now focused on what implementation breakthroughs this paper may deliver. A detailed examination of the paper will have to wait for further community discussion.

"AI Anxiety" and Burnout Spread Among Silicon Valley Tech Workers

According to Bloomberg, anxiety that AI could take their own jobs is spreading among tech workers in Silicon Valley, and burnout is becoming increasingly severe. The topic was also picked up on Hacker News, where it struck a chord.

The rapid evolution of AI tools has created a sense of uncertainty for programmers, designers, and other professionals — never knowing "when they might be replaced" — which is taking a mental toll. While some find that leveraging AI tools boosts their productivity, the "fear of being replaced by AI" is adding new stress to the workplace. This is a structural problem for the industry as a whole, and one that individual mental health support alone is unlikely to solve.

---