Shareholders of Core Scientific just rejected a $9 billion exit. That is a vote of no confidence in the immediate liquidity event—and a massive bet on the board's ability to execute a technical pivot. The board responded by announcing a partnership with AMD to transform Bitcoin mining infrastructure into AI data centers. The problem? The technical details are missing.
Context: Core Scientific is a Nasdaq-listed Bitcoin mining and AI hosting company (CORZ). It emerged from bankruptcy in 2023 with a restructured balance sheet. The company operates physical infrastructure: power substations, cooling systems, and ASIC racks. The new narrative is that these assets can be repurposed for GPU-based AI compute. The partnership with AMD is the strategic vehicle for this transformation.
But the market should audit the promise, not the poster. The AMD partnership announcement contains zero technical specifics: no megawatts committed, no Instinct GPU model confirmed, no software stack validation, no revenue split. From my experience auditing infrastructure projects, this is a classic pattern—a strategic press release designed to influence valuation without providing verifiable data.
Core: Systematic Teardown
The conversion from ASIC mining to GPU AI hosting is an engineering challenge, not just a financial one. Bitcoin miners operate on low-latency, high-reliability power but with minimal networking requirements. AI data centers require high-density liquid cooling, InfiniBand or RoCE networking, and GPU cluster scheduling. AMD’s ROCm software ecosystem is still catching up to Nvidia’s CUDA—a proven gap in production workloads. Core Scientific has not demonstrated any capability in these areas.
Forensics don't lie. The company’s past AI hosting contracts, like the one with CoreWeave, were for GPU hosting, but that was a separate facility, not a retrofitted mining site. The conversion economics are also uncertain. Bitcoin miners lock in low power prices via long-term power purchase agreements (PPAs). These are valuable for GPU compute, but the capital expenditure for retrofitting is high—estimated at $3-5 million per megawatt for liquid cooling and networking. The source analysis notes that no technical milestones or delivery data were disclosed. This is a red flag.
High yield is a warning, not a welcome. The rejected $9 billion acquisition sets a valuation anchor. Shareholders are effectively saying the company is worth more than $9 billion under the new strategy. But to achieve that, Core Scientific will likely need to raise capital—either through equity dilution or debt. The company’s post-bankruptcy debt structure is opaque, and any new financing will be costly given the risk profile.
Additionally, the AMD partnership may be more about AMD needing deployment sites than Core Scientific needing chips. AMD is fighting for GPU market share and needs real-world validation of its Instinct series. Core Scientific provides a testbed with low-cost power. But this is a symbiotic relationship, not a one-sided benefit. The value capture for Core Scientific depends on the terms of the contract—none of which were disclosed.
Contrarian: What Bulls Got Right
To be fair, the bulls have a point. Core Scientific’s core asset—access to cheap, stranded power—is genuinely valuable for AI compute. The AI industry is facing a power bottleneck, and companies like Core Scientific that own substations and PPAs have a structural advantage. The AMD partnership also diversifies the chip supply chain, reducing dependency on Nvidia. If AMD’s MI300 series gains traction, Core Scientific could be an early access partner.
But the contrarian view has a narrow window. The company must prove it can deliver reliable, high-performance GPU clusters at scale. The first data points will come from the next quarterly report. If they report actual megawatts deployed, utilization rates, and customer commitments, the bet might pay off. Until then, it is a narrative supported by zero data.
Code does not lie; people do. In this case, the code is the infrastructure performance data. Core Scientific has not provided any.
Takeaway: Accountability Call
Core Scientific’s shareholders just placed a high-risk bet. The board bet on a technical transformation that has no publicly verifiable milestones. The data will tell if they were right. I am watching for the first MW delivery and utilization rates. Anything less than detailed quarterly reports is a red flag. The clock is ticking—and the market will demand proof.