The arithmetic is stark. Core Scientific shareholders rejected a $9 billion acquisition offer, then immediately announced a strategic partnership with AMD. On the surface, this signals management’s confidence in a higher intrinsic value. But as a due diligence analyst who has spent years dissecting infrastructure plays, I see a different equation: the rejection of a concrete exit for a vague narrative of AI transformation. The market cheered the AMD news, but the underlying architecture reveals a critical debt in operational transparency.
Let me state the premise clearly. Core Scientific is a Nasdaq-listed entity (CORZ) that emerged from Chapter 11 bankruptcy in early 2024. Its core business was Bitcoin mining, but it has pivoted to hosting high-performance computing (HPC) and AI workloads. The $9 billion offer—widely attributed to CoreWeave, a competitor with whom Core Scientific already has a multi-year hosting contract—was rejected. The AMD partnership was presented as the justification. But the announcement lacked any technical specifics: no MW capacity committed, no GPU deployment timeline, no software stack validation. Code compiles, but context reveals the exploit.
Context: The Infrastructure Layer’s Identity Crisis
Core Scientific sits at the intersection of two capital-intensive industries: Bitcoin mining and AI cloud. Its competitive advantage is supposed to be its existing power infrastructure—long-term power purchase agreements (PPAs) that lock in low electricity costs. But converting a Bitcoin mining facility into an AI data center is not a simple retrofit. It requires liquid cooling, high-density racks, InfiniBand or RoCE networking, and a GPU cluster scheduler. The facility that once housed ASICs (Application-Specific Integrated Circuits) must be rewired for general-purpose GPUs. The engineering challenge is non-trivial, and the AMD partnership introduces an additional layer of complexity.
AMD’s Instinct GPUs target AI workloads, but their software ecosystem, ROCm, still lags behind Nvidia’s CUDA in maturity and developer adoption. Based on my experience auditing DeFi protocols in 2020, I learned that ‘compatibility’ is often a marketing term until proven under load. Core Scientific’s announcement contained no benchmark results, no reference architecture, no third-party validation. This is a strategic announcement, not a technical one. In the world of high-stakes infrastructure, that distinction is lethal.
Core: A Systematic Teardown of the AMD Partnership
The core of the thesis is that Core Scientific can replicate the success of its CoreWeave hosting contract with AMD. But let’s dissect the variables.
First, the revenue model. The CoreWeave contract—signed in 2024—is a hosting agreement where Core Scientific provides power, cooling, and physical security. The revenue is likely fixed per MW, with a term of 3-5 years. The AMD partnership, as described, could be a hardware procurement agreement, not a revenue-sharing deal. Without a committed offtake from AMD or its customers, Core Scientific is taking on inventory risk. If AMD’s chips face supply chain issues or performance disappointments, the facility sits idle. In 2021, I traced 15% of Bored Ape Yacht Club volume to wash trading; today, I see a similar pattern of inflated expectations masking real economic exposure. The Wash Trading Index applies here: the volume of announcements does not equal the volume of deployable compute.
Second, the technical debt. Bitcoin mining facilities are designed for continuous, low-maintenance operation. ASICs are rugged, with minimal networking requirements. AI data centers are the opposite: they require high-bandwidth, low-latency interconnects, constant software updates, and rigorous thermal management. The transition is not just a CapEx problem—it’s an OpEx problem. Staff must be retrained; SLAs must be rewritten. In my 2022 analysis of Terra/Luna, I compared Frax’s partial collateralization to Terra’s algorithmic model. The lesson was that market confidence is not a substitute for hard assets. Here, the confidence in Core Scientific’s ability to execute the pivot is not backed by a track record in AI hosting. The company’s history is mining, not deep learning.
Third, the capital structure. Core Scientific emerged from bankruptcy with a cleaned-up balance sheet, but the AI pivot requires massive capital expenditure. The estimated cost to convert a 100 MW mining facility to AI hosting can exceed $50 million, depending on density. Where does this capital come from? The company could issue new equity, diluting existing shareholders, or take on debt at a time when interest rates remain elevated. The rejection of the $9 billion offer implies that management believes the company is worth more. But that belief must be backed by a credible path to positive free cash flow. Based on my 2025 compliance audit experience, I know that regulatory requirements for AI data centers are also tightening—especially in the EU with MiCA and the upcoming AI Act. Core Scientific’s Portuguese operations (if any) would face additional scrutiny. The cost of compliance is often underestimated.
Contrarian: What the Bulls Got Right
To be fair, the bulls have a point. The AMD partnership is a strategic hedge against Nvidia’s dominance. If AMD’s MI300X or future chips gain traction, Core Scientific could offer competitive pricing and avoid single-vendor lock-in. The company’s existing power contracts are a genuine moat; in a world where AI compute demand is outstripping supply, access to cheap power is a scarce resource. The rejection of the $9 billion offer may also be a signal that the board believes the company’s AI hosting pipeline is worth more than the market currently prices. In 2020, when I published my report on Aave’s unsustainable yields, I was ridiculed—until the protocol paused minting. The contrarian view here is that the market may be underestimating the value of power infrastructure in a supply-constrained AI market.
However, the bulls are ignoring the execution risk. Core Scientific has never operated a large-scale AI data center. The AMD partnership, while promising, is not a guarantee of performance. The company’s stock price reacted positively to the news, but that reaction is based on narrative, not data. In my experience, narrative-driven rallies are the first to reverse when the first quarterly earnings miss expectations. The $9 billion offer set a floor; the AMD partnership sets a ceiling of expectations. The gap between the two is where the risk lies.
Takeaway: The Accountability Call
Core Scientific’s shareholders have made a high-stakes bet. They rejected a sure $9 billion for an uncertain future. The AMD partnership is a necessary step, but it is not a sufficient one. The company must now deliver operational milestones: MW of compute deployed, utilization rates, and revenue per MW. Without these metrics, the announcement is just noise. As I wrote in my 2025 compliance framework, ‘Verify. Then trust. Never assume.’ The market is now in a verification phase for Core Scientific. The next 12 months will reveal whether the rejection was a stroke of genius or a catastrophic miscalculation. Either way, the forensics are just beginning.
