CxO · Daily · May 28, 2026

the ai search backlash

Google's AI-heavy search rollout is driving measurable user flight. DuckDuckGo visits jumped 28% after Google's AI mode launch.[1] DuckDuckGo installs rose 30% as users reacted to AI summaries crowding out classic results.[2]

Users describe Google's AI mode as a chatbot that downgrades relevance, obscures diverse sources, and makes search less predictable, pushing them toward alternatives like DuckDuckGo and Brave despite occasional quality gaps.[3]

Self-hosters frame their shift as about control more than privacy, reinforcing a broader move away from default reliance on Google for discovery.[4]

ai infra costs and data center politics

Major companies are quietly throttling AI usage because running current models can cost more than human workers.[5] Uber burned through its entire 2026 AI budget in four months on Claude Code, even as leadership admits productivity gains are not yet showing up in company-level results.[6]

At the same time, investors are staring at projections of $725B in AI infrastructure spend by 2026.[7] AWS has a $6B chip deal with Snowflake, and Nvidia plans to invest $150B per year in Taiwan, locking even more capital into the stack.[8]

On the cost side, new inference frameworks claim 99% reductions in caching costs and there is rising interest in decentralized or self-hosted inference to escape cloud pricing, but these are early and unevenly adopted.[9]

Data centers are becoming political flashpoints: Lombardy slapped a 200% tax on builds in agricultural zones, local power bills near centers are reported at 250% above average, and Vancouver is already protesting planned AI facilities over water and energy use.[10]

agents on real money rails

Robinhood has shipped a credit card explicitly designed for AI agents, offering 3% cash back and letting those agents initiate purchases and stock trades.[11]

Users remain fully liable for agent actions, and there are effectively no bespoke rules yet on how far trading authority for autonomous agents can go, turning this into a live regulatory stress test.[12]

Parallel experiments like Rentahuman, which lets AI agents pay humans for real-world tasks, show the same pattern of agents moving from toy domains into labor and payments flows.[13]

The security stack is lagging: a critical vulnerability in the open-source Starlette framework allegedly exposed millions of AI agents, and practitioners say operational reliability of coding agents now matters more than raw model quality.[14]

Regulators are starting to catch up at the edges—the EU AI Act will cover autonomous agents by 2026–27, and Google is pitching AI Threat Defense while GPT-5.5 and similar models are touted for security use cases—but these tools coexist with rising FBI concern about anti-tech extremism rather than a coherent agentic risk regime.[15]

techno-nationalism and data sovereignty

Europe and its neighbors are translating data-sovereignty rhetoric into hard constraints. The EU AI Act sets binding rules for high-risk systems and autonomous agents by 2026–27, while many in the bloc explicitly oppose using non-EU providers for sensitive data after repeated privacy scares.[16]

The Netherlands blocked a U.S. acquisition of DigiD operator Solvinity on national-interest and data-sharing grounds, and Canadian defense just ordered a fleet of Swedish aircraft instead of relying on U.S. suppliers, both interpreted locally as moves away from default U.S. dependence.[17]

Within Europe, regulators favor European satellite services over Starlink and struggle over whether to ban Huawei, underscoring a tilt toward regional tech champions rather than a unified Atlantic approach.[18]

Talk of boycotting U.S. goods and distrust of U.S. tech firms on privacy are increasingly common in European discourse, especially after perceived overreach by American platforms and politicians.[19]

ai, jobs, and political blowback

An MIT report argues AI is not the main driver of recent tech layoffs, and Altman plus other lab leaders are now publicly downplaying earlier 'jobs apocalypse' rhetoric.[20]

Despite that, a global survey shows the share of CEOs planning to cut junior roles due to AI has jumped from 17% to 43%, and many firms are explicitly using AI as cover for broader restructuring.[21]

This gap is fueling labor and political backlash: NYT tech unions are filing complaints over AI monitoring tools, U.S. students are booing pro-AI speakers, and the FBI now treats anti-tech extremism as a domestic threat vector.[22]

Policy responses are moving from op-ed to statute, including Elizabeth Warren’s call to tax AI, Italy’s 200% tax on data centers in agricultural zones, and proposals to tax automation to rebalance incentives between labor and capital.[23]

At the same time, the Vatican’s 150-page Magnifica Humanitas encyclical—co-authored by an Anthropic founder—explicitly calls for 'disarming' AI and breaking big-tech monopolies, even as OpenAI sets up a $250M worker-adjustment fund and Nvidia’s Jensen Huang attacks CEOs for scapegoating AI in layoffs.[24]