PLUS: The AI subscription time bomb and Europe's two-year warning
Happy reading
Global advertising giant Publicis Groupe is making a massive $2.5 billion investment, signaling a major shift in AI strategy. The acquisition of data platform LiveRamp is a huge bet on the future of autonomous AI agents.
The deal puts the spotlight on the critical need for unique, proprietary data to power effective AI. As the industry moves from focusing on models to the data that fuels them, is owning the best dataset now more important than building the biggest model?
In today’s Next in AI:
Publicis Groupe’s $2.5B AI agent bet
The AI subscription time bomb
Europe’s two-year AI warning
The mainstream AI backlash
Ad Giant's $2.5B AI Agent Bet
Next in AI: Global advertising leader Publicis Groupe announced a landmark $2.5 billion deal to acquire data platform LiveRamp. The move aims to build the critical data foundation for the next wave of autonomous AI agents.
Explained:
Publicis is betting on a concept it calls "data co-creation," where companies securely combine different data sources to generate new, proprietary assets for training AI.
The acquisition builds on Publicis's 2019 purchase of Epsilon, layering LiveRamp's secure data-sharing technology on top of an existing identity graph with 2.3 billion global profiles.
LiveRamp was already pivoting heavily into AI, recently launching agentic orchestration and reporting strong FY26 revenue of $813M just before the deal was announced.
Why It Matters: This deal signals a major shift from focusing only on AI models to prioritizing the unique data that powers them. For businesses, it highlights that owning and combining proprietary data is becoming the ultimate competitive edge in deploying effective AI agents.
The AI Subscription Time Bomb
Next in AI: Major AI labs are subsidizing enterprise usage at a significant loss to win market share, creating artificially low prices. This sets the stage for a massive price correction that could blindside businesses that have built critical workflows on top of these temporary deals.
Explained:
The subscription math doesn't add up, with a standard $20/month seat often consuming between $200 and $400 in actual compute costs—a gap labs are currently absorbing.
Agentic AI is the catalyst breaking this model, with GitHub now moving to usage-based billing because autonomous agents made its flat-fee pricing unsustainable.
The exposure for businesses is immense, with U.S. companies projecting average AI spending of $207 million while embedding these underpriced tools deep into their core operations.
Why It Matters: Thousands of companies now depend on tools they believe are permanently cheap, creating a serious financial risk. As AI providers march toward IPOs and prioritize profitability, leaders must audit their true AI consumption now to avoid a budget-breaking shock later.
Europe's 2-Year AI Warning
Next in AI: Mistral AI’s CEO, Arthur Mensch, issued a stark warning to French lawmakers, stating Europe has just two years to build its own sovereign AI infrastructure before becoming permanently dependent on U.S. tech giants.
Explained:
Mensch’s warning goes beyond algorithms, focusing on the physical supply chain of chips, energy, and compute capacity that underpins AI dominance.
The company is backing its words with action, partnering with state-backed institutions to strengthen Europe's "digital sovereignty" through new GPU infrastructure.
To put the challenge in perspective, Mistral aims to build a gigawatt of AI computing capacity by 2029, as U.S. tech companies reportedly deploy a trillion dollars toward infrastructure.
Why It Matters: The race for AI supremacy is increasingly about controlling the physical infrastructure, framing it as a critical issue of geopolitical and economic sovereignty. This push for European independence could spark significant investment and create new opportunities in the continent's tech ecosystem.
The AI Backlash Goes Mainstream
Next in AI: Public sentiment is turning against AI, with a new wave of skepticism creating real-world roadblocks for the industry. From viral boos at commencement speeches to canceled data centers, the backlash is no longer just online chatter.
Explained:
Recent polls show a significant shift in public opinion, with one Economist/YouGov poll revealing that over 70% of Americans believe AI is advancing too quickly.
This growing resistance is hitting the industry's infrastructure, as a record number of AI data centers were canceled in early 2026 due to community pushback.
The sentiment is especially strong among younger generations, with tech executives getting booed for pro-AI remarks and a Gallup survey finding only 18% of young people feel hopeful about the technology's future.
Why It Matters: The AI industry faces a serious public relations problem that now directly threatens its growth trajectory. This public skepticism challenges the narrative of AI's inevitability and could slow the pace of innovation and adoption.
AI Pulse
Daring Fireball argues that AI is a foundational technology like wireless networking, not a product, pushing back on the narrative that Apple needs a singular "killer AI device" to compete.
Inc. suggests that the biggest hurdle for AI wearables isn't the tech itself, but passing the "coffee shop test"—the social awkwardness and public embarrassment of using unfamiliar gadgets around other people.
Frederick Vanbrabant explains why simply "throwing AI at it" won't speed up business processes, arguing that true acceleration comes from fixing upstream problems and providing high-quality, predictable inputs to bottlenecks.