AI week in focus: bubble risks, lighter rules in Europe, and Nokia's big reorg
Three signals this week matter for strategy leaders: the Google chief flagged froth in AI investing, Brussels eased the near-term compliance runway, and Nokia tightened its structure to chase AI infrastructure demand. Read this as a prompt to stress test budgets, rebalance risk, and reassess vendor roadmaps.
Google CEO warns of AI "irrationality" as valuations surge
Sundar Pichai said parts of the AI market look overheated and could "overshoot." He also pointed to rising energy demand straining current systems, potentially slowing the company's 2030 net-zero path, and reminded users that AI is still error-prone.
Context: Alphabet crossed $3 trillion earlier this year and Nvidia touched $5 trillion last month. Even long-term bulls are hedging; as one analyst put it, even Sam Altman has hinted we might be in a bubble.
Why it matters for executives
- Reprice risk: Assume a valuation reset scenario and test AI program ROI at higher energy and compute costs.
- Energy is strategy: Model data center power availability, tariffs, and sustainability constraints into AI capacity planning.
- Quality guardrails: Ship AI features with human-in-the-loop checks and clear error budgets.
Immediate actions
- Set hurdle rates for AI bets that include energy, retraining, and hallucination mitigation costs.
- Negotiate flexible cloud/GPU terms tied to utilization, not just reserved capacity.
- Establish incident reporting for AI output errors similar to SRE postmortems.
EU eases near-term AI compliance to spur innovation
The European Commission proposed delaying some high-risk AI Act requirements by up to 16 months, part of a broader push to cut red tape and improve competitiveness. The package also considers a GDPR update for model training data, lighter cookie consent, a unified AI reporting interface, and a European Business Wallet to help companies scale across member states. The EC estimates up to €5 billion in admin cost savings by 2029.
European Commission AI policy overview
Why it matters for executives
- More runway: Extra time to pilot AI in production, but don't confuse delays with deregulation-documentation discipline will still pay off.
- Data strategy shift: If GDPR rules loosen for training data, revisit consent, retention, and vendor contracts to avoid retroactive exposure.
- Operational leverage: A unified reporting interface could cut compliance overhead-design for it now to avoid rework.
Immediate actions
- Stand up a minimal AI compliance stack: model registry, data lineage, evaluation reports, and risk logs.
- Run a privacy impact review on any plan to use personal data for training or fine-tuning.
- Ask vendors how they will support the EU's unified reporting and wallet initiatives in their roadmaps.
Nokia reorganizes to capture AI-infrastructure spend
Starting 1 January 2026, Nokia will consolidate into two segments: Network Infrastructure (Optical, IP, Fixed-building on its $1B AI infrastructure deal with Nvidia) and Mobile Infrastructure (Core, Radio, Technology Standards) with a long-term focus on AI-native networks and 6G. Leadership will shift, including exits and new appointments in standards. Shares fell up to 6% on the news, and some analysts argue the direction isn't a radical break-but the capex and payoff timing are still uncertain.
Why it matters for executives
- Vendor posture: Treat network OEMs as AI infrastructure partners, not just box suppliers-expect bundled compute, acceleration, and automation.
- Procurement leverage: A simpler structure can streamline deals across optical/IP/fixed, but watch for lock-in around AI toolchains.
- Timing risk: 6G and AI-native network ROI will lag-stage investments to measurable automation wins in the near term.
Immediate actions
- Map workloads: Which AI functions belong at the edge, in the RAN/Core, or centralized data centers? Buy accordingly.
- Demand open interfaces and exportable telemetry for AI ops-no black boxes.
- Run a 12-18 month automation roadmap (fault prediction, energy optimization, self-healing) with clear savings targets.
What to do in the next 90 days
- Scenario plan for a GPU and power crunch. Prioritize use cases by margin impact and energy intensity.
- Set a portfolio rule: 70% on near-term automation, 20% on enabling platforms, 10% on moonshots. Review quarterly.
- Stand up an AI risk and reliability review before any external launch-factsheets, eval metrics, red-team results.
- Align contracts to volatility: include price-protection, performance SLAs, and exit ramps for AI services.
If you're upskilling leaders and teams to execute this plan, explore role-based programs at Complete AI Training for a practical curriculum you can roll out fast.
Your membership also unlocks:
