AI Bubble Trouble?

Core Scientific Shareholders Reject CoreWeave's $9B Stock Deal - Betting on Bigger AI Upside

Core Scientific shareholders voted down a $9 billion all-stock acquisition offer from partner and competitor CoreWeave. The decision followed a vote-no push from the company's largest shareholder, Sina Toussi of Two Seas Capital, who focuses on post-bankruptcy situations. Core Scientific exited bankruptcy in January 2024.

Both firms started as crypto miners. CoreWeave has since pivoted into AI infrastructure with investor and partner Nvidia, spending its stock on acquisitions as its valuation climbed from IPO to roughly $66 billion today (about $140 per share). Core Scientific still mines crypto, but it's been positioning its data centers for AI compute.

Before the buyout proposal, CoreWeave locked in a $10 billion, 12-year agreement to run AI workloads in Core Scientific facilities. The acquisition offer, announced in July, came at a premium to Core Scientific's price at the time. Toussi argued Core Scientific could grow into a CoreWeave-like asset on its own. His stance: with AI infrastructure investment accelerating, $16.40 per share wasn't compelling.

Shareholders voted no. CoreWeave walked. Core Scientific's stock jumped and now trades around a $6.6 billion market cap.

Why this matters

Investors turning down premium bids suggests confidence that AI-driven valuations can go higher. That's bullish - and another sign we may be nearing an AI bubble. When targets reject stock deals, it's often because they believe their own multiple is headed up faster than the buyer's.

What finance, IT, and developers should watch

• Investors: Expect more contested deals and partnership-over-ownership plays. If AI capacity remains scarce, data center operators with power and land can command better terms. But watch the spread between long-term contracts and takeout prices - it signals where smart money thinks intrinsic value is heading.
• IT leaders: The $10B, 12-year contract signals durable demand for AI compute. If you're building AI workloads, securing capacity (power, GPUs, cooling) and multi-vendor optionality is becoming a strategic hedge.
• Developers: CoreWeave acquired Marimo, an open source alternative to Jupyter Notebooks, to move up the stack toward app development. Notebooks are a core workflow for data and AI teams - if you use them, keep an eye on how hosted platforms integrate these tools for collaboration and deployment. Learn more about notebooks at Jupyter.

The strategic read

Core Scientific has a long runway if it converts mining-heavy assets into AI-ready capacity. Shareholders think that optionality is worth more than the current bid. CoreWeave, meanwhile, is buying tooling (like Marimo) to capture developer mindshare - not just GPU hours.

If you lead budgets or teams, the playbook is simple: secure compute, reduce vendor lock-ins, and build internal competency around AI workloads and tooling. The companies doing that now will dictate pricing and pace later.

Practical next steps

• Finance: Track GPU supply, data center power deals, and long-term capacity contracts - these are the new fundamentals for AI infrastructure plays.
• IT/Engineering: Pilot workloads across multiple providers to benchmark throughput, networking, and cost per token/epoch - not just list price per GPU.
• Developers: Standardize on notebooks and experiment tracking you can export or self-host, so you're not boxed in by a single platform.

If you're upskilling teams for AI projects, browse role-based learning paths here: AI courses by job.

Also worth watching: Nvidia's data center ecosystem to see where compute and tooling are consolidating. Start with their data center overview here: Nvidia Data Center.