Sunday, May 31, 2026
Sparked Daily — 2026-05-31 | AI Briefing for Founders & Leaders
1️⃣SoftBank Commits €75B to French Data Centers
SoftBank announced plans to invest up to €75 billion to build and operate 5 gigawatts of additional data center capacity in France. This represents one of the largest single infrastructure investments in European AI infrastructure to date.
Why it matters: Europe is about to get a massive AI infrastructure boost that could shift the global balance of compute power. For AI companies struggling with GPU access, this creates a new major alternative to US cloud providers. French startups and European enterprises suddenly have a path to compete with Silicon Valley on infrastructure terms. Watch for this to trigger similar mega-investments across Europe as countries realize data sovereignty isn't just about regulation — it's about having the physical infrastructure to train and run competitive AI models.
2️⃣GitHub Copilot's Token Billing Sparks Developer Revolt
Microsoft is shifting GitHub Copilot from flat-rate pricing to token-based billing, triggering widespread developer backlash. The change effectively makes heavy users pay significantly more for the same service they previously had unlimited access to.
Why it matters: This is Microsoft testing how much developers will pay for AI coding assistance — and the answer appears to be 'not as much as we hoped.' The backlash signals that AI coding tools haven't yet proven indispensable enough to command premium pricing. For competing tools like Cursor, Codeium, and Replit, this is a massive opportunity to win over price-sensitive developers. Expect a race to the bottom on AI coding tool pricing, which ironically might accelerate adoption by making these tools accessible to smaller teams and individual developers.
3️⃣Meta Reportedly Building AI Pendant for Wearables
Meta is developing an AI-powered pendant as part of its broader push into AI hardware beyond VR headsets. The device would join the company's Ray-Ban smart glasses in creating an ecosystem of always-on AI assistants.
Why it matters: Meta is betting that AI will make wearables mainstream in ways that fitness trackers and smartwatches couldn't. A pendant form factor suggests they're targeting a different market than Apple Watch — something more subtle and always-accessible for AI interactions. If successful, this could establish Meta as the leader in ambient AI interfaces before Apple figures out its own strategy. The real test will be whether people want AI assistants that are literally always with them, or if that crosses into creepy surveillance territory.
4️⃣AI Chip Startup Groq Raising $650M
AI inference chip maker Groq is reportedly raising $650 million as it pivots from pure hardware to focus more on AI inference services. This follows Nvidia's recent $20 billion acquisition spree in the chip sector.
Why it matters: Groq's massive fundraise signals that specialized AI inference chips are becoming a legitimate alternative to Nvidia's training-focused GPUs. For enterprises running AI models in production, this could mean dramatically lower costs and better performance for inference workloads. The pivot to services suggests even hardware companies realize the real money is in the software layer. If Groq can deliver on speed and cost advantages, it could crack open Nvidia's inference monopoly just as demand for running AI models explodes beyond training into real-time applications.
5️⃣Box CEO Aaron Levie Diagnoses 'AI Psychosis'
Box founder Aaron Levie coined the term 'AI psychosis' to describe executives who don't understand jobs they're replacing with AI. This comes as ClickUp cut 22% of its workforce for AI agents and 2026 tech layoffs nearly match all of 2025.
Why it matters: Levie just named the biggest risk in AI adoption: leaders making decisions about AI capabilities without understanding the work being automated. ClickUp's massive layoffs for AI agents are likely the canary in the coal mine — expect more companies to cut staff prematurely, then struggle when their AI can't handle edge cases and complex judgment calls. Smart founders should see this as a competitive advantage: while competitors cripple themselves with AI psychosis, you can use AI to augment rather than replace key talent. The companies that get this balance right will dominate the ones that don't.
⚡ Spark's Take
The Great AI Reality Check: When Hype Meets Hard Economics
The AI boom is entering its awkward teenage phase — old enough to make real money, young enough to make catastrophic mistakes. Today's stories reveal an industry grappling with the gap between AI's promise and its practical limitations, from SoftBank betting €75 billion on European data centers to developers revolting against GitHub's pricing changes to executives suffering from what Box's Aaron Levie brilliantly dubbed "AI psychosis."
The common thread? We're moving from "AI can do anything" to "AI can do specific things really well, but it'll cost you."
1. SoftBank Commits €75B to French Data Centers
SoftBank just announced the largest single AI infrastructure bet in European history: €75 billion to build 5 gigawatts of data center capacity in France. To put that in perspective, that's enough compute to power a small country's entire AI ambitions.
This isn't just about data centers — it's about data sovereignty becoming data supremacy. European companies have been stuck choosing between expensive, limited local compute or sending their data to US cloud providers. SoftBank's investment creates a third option: European-based infrastructure with the scale to compete globally.
For AI startups, this could be transformative. Training large models currently means either paying premium prices to AWS/Google/Azure or waiting months for GPU access. A new European compute powerhouse could break that bottleneck, especially for companies handling sensitive data that can't leave the continent.
🔥 Spark's Hot Take: This move will trigger a global infrastructure arms race. Watch for similar mega-investments from sovereign wealth funds across the Middle East and Asia. The country that builds the most compute capacity in the next three years will have a permanent advantage in the AI economy.
2. GitHub Copilot's Token Billing Sparks Developer Revolt
Microsoft just learned that developers have strong opinions about pricing changes. The shift from unlimited GitHub Copilot usage to token-based billing has triggered a full-scale revolt, with developers calling the move "a joke" and threatening to switch tools.
The math is brutal: heavy Copilot users who previously paid $10/month for unlimited access could see bills jump to hundreds of dollars monthly. Microsoft essentially took their most engaged users and gave them the biggest price increases.
But here's what makes this fascinating — it reveals that AI coding tools haven't yet become indispensable. If Copilot were truly irreplaceable, developers would grudgingly pay the higher prices. Instead, they're shopping around, which suggests the moat around AI coding assistants is shallower than anyone expected.
Cursor, Codeium, Replit, and other competitors must be celebrating. Microsoft just handed them their best marketing campaign: "We won't suddenly jack up your prices." Expect a race to the bottom on AI coding tool pricing, which might actually accelerate adoption by making these tools accessible to individual developers and smaller teams.
3. Meta Reportedly Building AI Pendant for Wearables
Meta is betting that the future of AI isn't in your pocket — it's around your neck. The company is reportedly developing an AI-powered pendant to complement its Ray-Ban smart glasses, creating an ecosystem of always-on AI assistants.
The pendant form factor is clever. Unlike smartwatches that require you to raise your wrist or phones that need to be pulled out, a pendant is always accessible for quick AI interactions. It's also more socially acceptable than glasses for people who don't normally wear them.
This represents Meta's broader thesis that AI will make wearables mainstream in ways that fitness tracking and notifications never could. Instead of checking your heart rate, you're having conversations with an AI that knows your context, schedule, and preferences.
🔥 Spark's Hot Take: Meta is building the AI equivalent of AirPods — a device category that didn't exist until Apple created massive demand overnight. If they nail the user experience, pendant-style AI assistants could become as ubiquitous as wireless earbuds within five years.
4. AI Chip Startup Groq Raising $650M
Groq's massive $650 million fundraise signals a fundamental shift in the AI chip market. While everyone obsesses over training models, Groq is betting the real money is in inference — actually running AI models in production.
The timing is perfect. As AI models move from research labs to real applications, companies need chips optimized for speed and efficiency rather than raw training power. Groq's architecture promises dramatically faster inference at lower costs than traditional GPUs.
Their pivot from pure hardware to inference services is telling. Even chip companies realize that software and services capture more value than silicon. If Groq can deliver on their performance promises, they could crack open Nvidia's inference monopoly just as demand explodes.
For enterprises, this could mean the difference between AI features that feel laggy and expensive versus ones that feel instant and affordable. That's the difference between AI being a nice-to-have feature and becoming core to your product experience.
5. Box CEO Aaron Levie Diagnoses 'AI Psychosis'
Box founder Aaron Levie just named the biggest risk in AI adoption: executives who don't understand the jobs they're automating. He calls it "AI psychosis" — the condition where decision-makers assume AI can replace human judgment without understanding what that judgment actually involves.
ClickUp's recent decision to cut 22% of its workforce in favor of AI agents is the perfect example. The executives making these cuts likely don't spend their days handling edge cases, managing complex customer relationships, or making nuanced decisions that AI can't yet handle.
This creates a massive competitive opportunity. While your competitors cripple themselves with premature AI replacements, you can use AI to augment your best people instead of replacing them. The companies that get this balance right will dominate the ones that don't.
The 2026 tech layoff numbers are already approaching 2025's total, and we're only in May. That's not just about economic uncertainty — it's about AI psychosis spreading through executive suites like a virus.
Bottom Line
The AI industry is growing up fast, learning hard lessons about the difference between demos and deployment, between hype and sustainable business models. SoftBank's massive European bet shows where the smart money is going (infrastructure), while GitHub's pricing revolt shows where it's not (squeezing early adopters). The winners will be companies that enhance human capability rather than hastily replacing it — because as Aaron Levie warns, AI psychosis is real, and it's expensive.
The question isn't whether AI will transform your industry — it's whether you'll be smart enough to avoid the expensive mistakes everyone else is about to make.
Want this in your inbox every morning?
Sign up free — 5 AI takeaways delivered before your morning coffee.