Marvell's $115B Lumentum deal reshapes AI chip competition

Political ad sentiment on data centers collapses to 3% support; Census Bureau tracks 'Super Intelligence' adoption at 24.6% of firms

Meta political ads mentioning data centers shifted dramatically from 42% support in early 2026 to just 3% support by October, with 57% now opposing them. This signals a major shift in public messaging around AI infrastructure that could affect permitting and buildout timelines for compute facilities. In parallel, the US Census Bureau's business survey now asks firms whether they use "Super Intelligence" (defined broadly to include voice recognition), revealing rapid enterprise adoption growth. The survey found 24.6% of firms reported using it in early October, up from 17.3% in late November 2025, providing an official government data signal on AI tool penetration across the US business base.

Grok Bot emerges as strongest personal AI agent in head-to-head benchmark against Claude, Dot, and Muse

In a nine-task real-world comparison, Grok Bot scored 93% overall and won on Figma design, inbox triage, and shopping tasks, making it the top choice for professional workflows. This signals clear product differentiation in the emerging personal agent category and suggests XAI's agent implementation is outpacing competitors on practical utility.

Figma's PM organization built a 31-skill AI system in Git to automate routine work and run continuous growth experiments

Figma's product management team deployed an AI system (PM OS) with agents for bug-bashing, Slack triage, experiment iteration, and real-time writing feedback, with a single human checkpoint before experiment launch. This demonstrates concrete enterprise AI adoption patterns and suggests design-tool companies are becoming platforms for agent deployment.

Why Marvell Is Buying Lumentum for $115 Billion

Data Gravity published an essay analyzing Marvell's $115B acquisition of Lumentum and its strategic implications for AI infrastructure. Main takeaways:

• Co-packaged optics (CPO) eliminates demand for Marvell's DSP chips but lasers remain essential; Lumentum's InP laser capacity is sold out through early 2029 with 70% of demand unmet
• Combined entity reaches ~$19B revenue at 40% operating margin ($7.6B operating income); with $1B cost-savings target, margins expand to ~45%
• Lumentum contributes 34% of revenue but 38% of operating profit, making all-stock deal accretive to Marvell EPS in year one despite 18x revenue multiple
• Broadcom loses exclusive full-stack advantage (accelerator, switch, DSP, optical engine, laser); Nvidia faces supply risk as Lumentum becomes competitor's subsidiary
• Marvell acquires five InP fabs, ending fabless model; capex increases and gross margins become volatile

On Facebook and Instagram, Political Ads Are Increasingly Critical of Data Centers

Pew Research Center published a data-driven analysis of political advertising trends around U.S. data centers ahead of the 2026 midterm elections. Main takeaways:

• Anti-data center political ads on Facebook and Instagram surged from 31% of data center-related ads in January 2026 to 57% in October, while pro-data center ads collapsed from 42% to 3%.
• $12 million was spent on data center-related political ads over 90 days (July 27–October 24, 2026), with Wisconsin, Ohio, and Texas accounting for the largest spending ($1.45M, $1.82M, and $1.69M respectively).
• Anti-data center ads disproportionately emphasize corporate interests (47% vs. 22% in pro-data center ads), while pro-data center ads focus on job creation (25% vs. 3%).
• Democratic and Republican candidates are similarly critical of data centers when they mention them, despite Democrats running more data center ads overall.

China's AI Jobs Problem: What Economists Say

Sinification published a report on Chinese economists' assessments of AI's employment impact. Main takeaways:

• State policy promotes AI diffusion and retraining for new roles (data annotation, AI training), but economists doubt fiscal capacity to fund large-scale retraining programs.
• Near-term consensus: AI adoption will severely displace workers and lower incomes, intensifying existing demand weakness and deflationary pressures rather than creating new opportunities.
• Demographic headwinds compound risks: aging population encourages automation while limiting service demand; graduate unemployment already elevated despite rising graduate numbers.
• Flexible/gig work (320M projected workers in 2026) positioned as employment buffer but risks creating lower-paid, less productive jobs; economists debate whether new AI-related roles will materialize.
• Proposed funding mechanisms (VAT-based social security, 'robot taxes' on monopoly profits, 'skills accounts') face uncertainty about adequacy and implementation timeline.