The Signal — May 29, 2026
The AI industry's public narrative is shifting fast. Executives are softening doomsday predictions just as their companies' spending reveals how deeply they've already committed to the technology, and the semiconductor world may have just gotten a new rulebook.
Sam Altman Reverses on AI "Jobs Apocalypse"
Earlier this week, OpenAI CEO Sam Altman walked back his earlier predictions that AI would eliminate most human jobs. Speaking at a Commonwealth Bank of Australia conference on Monday, Altman said: "I don't think we're going to have the kind of jobs apocalypse that some of the companies in our space advocate or talk about." That's a clear reversal from someone who previously warned of sweeping workforce displacement.
The timing is hard to ignore. Altman's softer tone arrived days after OpenAI filed its confidential IPO registration, a move that puts the company on a path toward public market scrutiny. OpenAI currently operates at a $1.22 loss for every dollar of revenue it generates, making a reassuring public posture essential. Anthropic CEO Dario Amodei has made a similar pivot, tempering his own apocalyptic rhetoric in recent weeks.
The pattern suggests a coordinated recalibration across frontier AI companies as they transition from venture-backed startups to publicly traded entities. Investors pricing IPOs want growth stories, not existential warnings. Watch whether this rhetorical shift persists through OpenAI's roadshow, or whether the old predictions quietly resurface once the stock is trading.
Sources: Reuters · Time · Yahoo Finance
Uber Burned Through $3.4B AI Budget in Four Months
Uber's CTO revealed the company exhausted its entire $3.4 billion 2026 AI budget before May, barely four months into the fiscal year. The blowout followed an aggressive internal push to deploy Claude Code across engineering teams, complete with leaderboards ranking departments by AI usage. What was meant to drive productivity became a runaway cost center.
Anthropic's shift from flat-fee to per-token billing triggered part of the budget explosion. Autonomous agent loops, left to iterate on complex tasks, consumed thousands of tokens per execution. Multiply that across an engineering organization incentivized to maximize usage, and the math gets ugly fast. COO Andrew Macdonald acknowledged the difficulty: "It's very hard to draw a line between AI usage and improved consumer features."
This is the first major case study in what happens when enterprise AI adoption outruns cost controls. Uber's experience suggests that usage-based AI pricing, combined with agentic workflows generating unpredictable token volumes, creates a different cost profile than traditional software licensing. Every company deploying AI coding agents at scale should be paying attention to how Uber navigates the rest of 2026 with no AI budget left.
Sources: Yahoo Finance · Project Flux
Huawei Unveils "Tau Scaling Law" as Post-Moore's Law Chip Strategy
At IEEE ISCAS 2026 in Shanghai this week, Huawei executive He Tingbo (widely known as the "Chip Queen") presented the Tau Scaling Law, a semiconductor scaling methodology that replaces the traditional geometric approach of shrinking transistors with a time-based performance improvement framework. The company disclosed that 381 chips have already been designed and mass-produced using this approach over six years.
Huawei claims the Tau Scaling Law can deliver chips equivalent to 1.4nm processes by 2031, potentially without access to the extreme ultraviolet lithography equipment that US export controls have blocked. If validated at scale, this represents a credible alternative path to advanced semiconductors, one born directly from the constraints imposed by sanctions.
The open question: can Tau scaling match the raw performance gains of conventional node shrinks, or does it produce "good enough" chips that serve China's domestic market while falling short of cutting-edge global competition? Either way, it challenges the assumption that export controls would permanently cap Chinese semiconductor capability. The rest of the industry now has to factor in a competitor playing by different physical rules.
Sources: Reuters · Wired · DigiTimes
On the Editor's Desk
Killed today: Microsoft's cancellation of Claude Code licenses was two weeks stale (May 14) and already widely covered. An NVIDIA executive's comment that compute costs now exceed employee costs at some companies was interesting but too thin to stand alone without additional reporting. Dario Amodei's softening on AI job displacement was folded into the Altman story as supporting context rather than run separately, since the underlying dynamic (frontier labs moderating rhetoric ahead of funding events) is the same story.
