Data Kimi Model
AI's Reckoning: Record Spending, Falling Stocks, and a Shrinking Talent Pipeline
Alphabet and Tesla both reported quarterly revenues that beat analyst estimates after hours on Wednesday — and both saw their share prices fall. The reaction captures the central tension in the AI economy: investors are looking past top-line numbers to capital expenditure figures and finding staggering infrastructure spending without returns at the scale that spending implies. Jamie Dimon has warned that financial returns to investors funding AI infrastructure will likely disappoint relative to current valuations, a prediction that appeared to play out in real time.
Tesla's full second-quarter picture is particularly striking. The company posted record quarterly revenue of 28.2 billion dollars while operating income fell 57 percent year-over-year and capital spending more than doubled. Volume production for both the Cybercab and the Semi has been pushed past 2026, meaning two of the company's most anticipated products will not generate meaningful revenue this year as previously expected. The National Highway Traffic Safety Administration has separately demanded that Tesla explain Elon Musk's social media posts apparently depicting him texting and drinking espresso while driving, as part of an ongoing Full Self-Driving probe.
OpenAI President Greg Brockman publicly called Moonshot AI's Kimi K3 model 'pretty good' — unusually generous praise directed at a direct competitor — while adding uncertainty about whether Kimi achieved its performance through knowledge distillation from models like GPT, which would constitute the kind of intellectual property concern the White House has flagged. The White House has accused Moonshot AI of IP theft specifically in the context of Chinese AI companies building competitive models. Both assessments can coexist: Kimi K3 may be genuinely impressive and may also have been built in part using unauthorized access to U.S. model outputs.
Anthropic held acquisition talks with Physical Intelligence, the robotics AI startup founded by former Google DeepMind and Google Brain researchers focused on training general-purpose robot AI. The talks did not produce a deal, reportedly due to disagreements over valuation or integration terms. That Anthropic explored the acquisition at all signals that large language model companies increasingly view physical-world AI interaction as the next major frontier. Amazon, meanwhile, is cutting jobs in its Artificial General Intelligence division — a possible signal that the company is pivoting toward AI deployment and cloud services rather than competing directly in frontier model development.
Perhaps the most counterintuitive data point in the AI boom: computer science enrollment at U.S. universities has fallen for the first time in roughly 20 years. The leading hypothesis is that the AI boom is paradoxically making students question whether a CS degree is worth the investment — if AI can write code, why spend four years learning to code? The calculation may be rational in the short term but shortsighted, given that building and maintaining AI systems still requires deep technical knowledge.