The shareholders of Core Scientific just rejected a $9 billion buyout. In a bull market, that’s unusual. Most public companies would take the cash and run. But here, the board decided to bet on an AMD partnership instead. The narrative is clear: pivot from Bitcoin mining to AI infrastructure. But I’m not here to debate valuations. I’m here to audit the code—or in this case, the lack of it.
Context: The Infrastructure Layer Core Scientific is a Nasdaq-listed company (CORZ). It operates Bitcoin mining facilities and, more recently, high-performance computing (HPC) data centers for AI workloads. This is not a smart contract protocol. It’s a physical infrastructure play: repurposing mining power plants for GPU clusters. The market is bullish on this narrative. The AMD deal is marketed as a strategic shift. But when I read the announcement, I saw a ghost in the audit: no technical details, no performance benchmarks, no delivery timelines. Trust is math, not magic. And here, the math is missing.
Core: The Silicon-Level Audit Let’s strip away the hype. The AMD partnership is a supply chain diversification move. Core Scientific is buying AMD Instinct GPUs to power its AI hosting services. On paper, this makes sense. Nvidia dominates the GPU market, and AMD offers a cheaper alternative. But the devil is in the implementation details—or the absence of them.
First, the technical feasibility. Converting a Bitcoin mining facility to an AI data center is not trivial. Bitcoin miners use ASICs, which are power-hungry but simple. AI workloads require GPUs, which need high-density rack configurations, liquid cooling, and low-latency networking like InfiniBand or RoCE. Core Scientific has experience with power infrastructure, but the networking and cooling stack is a different beast. I’ve seen similar transitions in other projects, and the failure rate is high. The setup costs are massive, and the engineering complexity often leads to delays.
Second, the AMD ecosystem. AMD’s ROCm software stack is not as mature as Nvidia’s CUDA. This is a well-known bottleneck. For AI inference, ROCm works, but for training, the performance gap is significant. Core Scientific will need to work with AMD on joint engineering optimization to bridge this gap. The announcement didn’t mention any such collaboration. The question is: will AMD provide on-site engineers to optimize the hardware? Or is this just a hardware purchase agreement? Based on my experience auditing ZK-rollup circuits, I’ve learned that theoretical performance claims rarely match real-world benchmarks. The same applies here.
Third, the missing metrics. The article didn’t disclose any megawatt capacity, utilization rates, or contract terms. This is a red flag. For an AI infrastructure company, the key performance indicator is deployed power capacity in MW, not the number of GPU units purchased. Without this data, the AMD deal is a headline, not a technical milestone.
Contrarian: The Hidden Blind Spots The contrarian angle here is that the AMD partnership might be a defensive move, not a strategic one. The shareholders rejected a $9 billion buyout, which implies they believe the company is worth more. But the AMD deal doesn’t immediately create that value. It’s a long-term bet that might pay off in three to five years, assuming no execution failures.
But there’s a deeper risk: supply chain concentration. If AMD’s Instinct GPU production is delayed or yields are lower than expected, Core Scientific’s capacity commitments will be compromised. I’ve seen this happen in the crypto mining space with ASIC manufacturers. The risk is not just technical; it’s operational. The company’s debt load from the 2023 bankruptcy restructuring adds another layer of fragility. The entire deal is a balancing act: high capital expenditure, uncertain revenue, and a tight timeline.
Another blind spot is the AI market itself. The demand for AI infrastructure is real, but it’s not infinite. If the market cools down, the value of these facilities drops. The 90 billion buyout valuation sets a floor, but it also sets a target. The company must deliver that value through actual operations, not just Partnerships.
Takeaway: The Vulnerability Forecast The Core Scientific case is a cautionary tale for the bull market. The AMD deal is a strategic move, but it’s not a technical breakthrough. The shareholders are betting on execution, not innovation. The real test will be in the next six months: will the company deliver on its MW targets? Will the ROCm ecosystem perform? Or will the market realize that the emperor has no clothes?
Silence speaks louder than the proof. The lack of technical details in the announcement is a warning sign. I’ll be watching the upcoming quarterly reports for actual deployment data. Until then, the AMD partnership is a promise, not a guarantee. The code is not yet written.