The Omniscient Bulletin · 2026-05-29
The Omniscient Bulletin — May 29, 2026
Capital flows one way; the data flows another.
The AI economy has a consistency problem. Investors are writing $26 billion checks for coding agents while researchers document developers refusing to work without AI tools that, by at least some measures, make their code worse. Both facts can be true simultaneously - and the dissonance is exactly where the next reckoning will come from.
Industry
OpenAI switches on CPA ads inside ChatGPT, targeting $102B ad revenue by 2030
OpenAI has quietly turned on cost-per-action advertising inside its ads manager, letting select advertisers pay only when a user converts - clicks through, signs up, or purchases. The move follows a conversion-tracking pixel OpenAI rolled out earlier this month and marks a pivot from brand-awareness spend toward performance advertising. With $14B in projected losses this year and an IPO expected before year-end, OpenAI is building an ad stack that competes directly with Meta and Google.
Agents
Cognition raises $1B at $26B valuation as Devin hits $492M annualized revenue
Cognition, maker of AI software engineer Devin, has closed a Series D of more than $1 billion at a $26 billion post-money valuation - more than double its $10.2 billion valuation from just eight months ago. Revenue has grown 13x year-over-year to $492 million ARR, with enterprise usage up 50% month-over-month for six consecutive months. Customers include Mercedes-Benz, Goldman Sachs, and Santander. The round was led by Lux Capital, General Catalyst, and 8VC.
Compute
Nvidia commits $6.5B to photonics firms in three months, betting on light to solve AI's energy bottleneck
Nvidia has poured at least $6.5 billion into photonics companies since March, including $2 billion each into Lumentum, Coherent, and Marvell, a $500 million investment in Corning, and participation in Ayar Labs' $500 million Series E. The strategy bets that optical data transfer - using light instead of copper - will be essential to scaling AI infrastructure without hitting energy and bandwidth walls. Jensen Huang said at GTC that silicon photonics capacity needs to grow "substantially" beyond what the world has today.
Industry
Glean triples to $300M ARR in 15 months, selling enterprises on AI token cost reduction
Glean has crossed $300 million in annualized revenue, up from $100 million 15 months ago, as its pitch shifts from pure AI search toward a cost-control angle. CEO Arvind Jain says Glean's context graph - which connects to and learns from internal enterprise systems - reduces the number of tokens an AI must consume per task. At a moment when AI compute bills are straining enterprise budgets, that efficiency argument is becoming a lead selling point. The company was last valued at $7.2 billion in June 2025.
Research
METR survey: technical workers claim 1.4-2x productivity gains from AI, but self-reports likely overstated
A METR survey of 349 technical workers found median self-reported value gains of 1.4-2x from AI tools, with speed gains reported at 3x. But METR flags serious caveats: a 2025 RCT found workers overestimated AI's time savings by 40 percentage points on average, and the survey's own staff - who are most familiar with that prior work - reported the lowest gains of any subgroup. The gap between perceived and measured productivity may be AI's most consequential open question.
Labor
Schneider Electric deploys AI across 160,000 workers with no layoffs - NYT profiles the augmentation playbook
The New York Times today profiles Schneider Electric's AI rollout as a counterpoint to the industry's default displacement narrative. The French industrial giant has integrated AI into manufacturing and knowledge work across its full workforce without cutting headcount, instead using the technology to reduce waste and expand output per worker. With 2026 tech layoffs approaching 150,000 globally, Schneider's approach is notable - though whether it scales beyond capital-intensive manufacturing remains an open question.