On a Tuesday that felt like any other in this bear market, Core Scientific's shareholders did something remarkable: they said no to nine billion dollars. The proposed acquisition, a lifeline for a company that emerged from bankruptcy just over a year ago, was rejected. Instead, the board announced a partnership with AMD to supply GPUs for AI workloads. The market cheered. The stock jumped. But I couldn't shake the feeling that we were all reading the wrong story.
This isn't just a corporate finance event. It's a referendum on how we value infrastructure in the age of AI and decentralization. And it reveals a dangerous gap between what we celebrate and what we should be measuring.
I've been in this space long enough to remember the ICO boom of 2017, when every white paper was a promise and every partnership was a moon shot. That era taught me a painful lesson: announcements are not deliveries. Core Scientific's shareholders are betting that the company can transform itself from a Bitcoin mining survivor into an AI computing powerhouse. But the details matter. And the details are missing.
Context: From Mining to AI, A Second Act
Core Scientific is not a startup. It's a publicly traded Bitcoin miner (NASDAQ: CORZ) that survived the 2022 bear market through Chapter 11 reorganization. That experience shaped its DNA. The company owns massive data centers built for energy-intensive mining—facilities with long-term power purchase agreements that lock in low electricity costs. These are the same assets that can host GPU clusters for AI training and inference.
This pivot is not unique. Riot Platforms, Marathon Digital, and others are exploring similar transitions. But Core Scientific has a head start: in 2024, it signed multi-year hosting contracts with CoreWeave, a pure-play AI cloud provider. The AMD partnership announced alongside the rejected sale is the next step.
— Root: The 2022 Bear Market. I remember the despair of that year, when we launched the Resilience Hub to help junior developers stay in the industry. The lesson was clear: survival requires adaptation, but adaptation without transparency is just speculation. Core Scientific's shareholders are betting on adaptation, but they are doing so without the data that would make that bet rational.
Core Insight: The Three Bottlenecks No One Is Talking About
Let's get technical. Converting a Bitcoin mining facility into an AI data center is not a simple plug-and-play operation. It requires solving three major engineering challenges:
- Cooling: Bitcoin miners are air-cooled; AI GPUs require liquid cooling or high-density immersion. Retrofitting existing infrastructure is expensive and time-consuming. Core Scientific has not disclosed any capital expenditure plans for this conversion.
- Networking: Mining operations use simple peer-to-peer communication; AI workloads require high-bandwidth, low-latency interconnects like InfiniBand or NVIDIA's NVLink. Building a cluster of thousands of AMD Instinct GPUs with ROCm software stack demands a network fabric that is radically different from what a mining farm uses.
- Software Stack: This is the most underestimated bottleneck. NVIDIA's CUDA ecosystem is the default for AI development. AMD's ROCm is improving, but it still lags in compatibility, tooling, and community support. Every AI startup I've spoken to in the past year tells me the same thing: they want to use AMD for cost savings, but they end up sticking with NVIDIA because the developer experience is smoother. This is a cultural and technical inertia that no partnership announcement can solve.
Code is law, but people are the protocol. The real value of the AMD partnership is not the hardware—it's the commitment to build a software ecosystem that can compete. But that commitment is measured in developer hours, not press releases.
I've audited enough smart contracts to know that the most dangerous risks are the ones hidden in plain sight. In this case, the risk is not that Core Scientific will fail to deliver AI capacity—it's that the market will assume success without demanding proof.
Contrarian Angle: The Tyranny of the Announcement
Here's the uncomfortable truth: the shareholders who rejected the $9 billion sale may have made a mistake. They are betting that the AMD partnership and the AI pivot will create more value than a guaranteed exit. But that bet requires Core Scientific to execute flawlessly in a market dominated by NVIDIA, Amazon, and Microsoft—companies with decades of infrastructure experience and unlimited capital.
Consider the alternatives. If Core Scientific had accepted the acquisition, its shareholders would have received a premium of roughly 30% above the pre-announcement stock price. By rejecting it, they are essentially saying: "We trust management to deliver more than that." But trust is not a governance mechanism. It's a feeling.
Governance isn't a token sale; it's a social contract. In the DAO world, we learned that voting is not the same as governing. Uninformed delegation leads to centralization. The same applies here: shareholders voted on a binary choice without the data needed to make an informed decision. Where is the contract with CoreWeave? What are the revenue projections? What is the utilization rate of the existing GPU capacity? None of this was disclosed.
My experience during DeFi Summer taught me that governance is most effective when it is informed. I led a team that audited Uniswap's early governance mechanisms, and we found that the most engaged voters were the ones who had access to transparent data. Without that data, voting becomes a popularity contest. Core Scientific's shareholders are voting on a vision, not a plan.
Takeaway: Watch the Watts, Not the Headlines
The AMD partnership is a strategic hedge, not a home run. It diversifies the supply chain away from NVIDIA, but it does not solve the fundamental challenge of building a competitive AI cloud. The real metrics to watch are not the stock price or the partnership announcements. They are:
- Megawatts delivered: How many MW of GPU capacity is actually online and generating revenue?
- Utilization rate: Are the GPUs being used for paid workloads, or are they sitting idle?
- Customer contracts: Are the customers real AI labs, or just other miners trying to pivot?
In the crypto world, we often mistake hype for substance. We did it with ICOs, with DeFi, with NFTs. Now we are doing it with AI infrastructure. The question is not whether Core Scientific can pivot—it's whether we are willing to wait for the data before we celebrate.
As I wrote in the 2026 Autonomous Agent Accountability Charter, the convergence of AI and crypto demands a new level of transparency. We cannot hold machines accountable if we don't hold ourselves accountable first. Core Scientific's shareholders have made a bold bet. Now it's time for the company to earn that trust with evidence, not announcements.
We didn't build decentralized networks to replace centralized bankers with centralized miners. The real promise of this industry is that we can build systems that are transparent, verifiable, and inclusive. Core Scientific has a chance to model that transparency. But the clock is ticking, and the bear market waits for no one.
— Root: The 2022 Bear Market, where we learned that survival is not enough. We need to thrive with integrity.