Shareholders just torched a $9 billion exit. The message is clear: they believe the company is worth more. But the market is pricing the AMD partnership as the catalyst. I see a different story—one buried in the infrastructure, not the press release.
Let me be direct. Core Scientific is a Bitcoin miner that nearly died in 2022. They emerged from bankruptcy with a plan: repurpose their power infrastructure for AI data center hosting. The AMD partnership is the headline. The reality is far more complex.
Context: The Infrastructure Pivot
Core Scientific operates at the intersection of two energy-intensive industries: Bitcoin mining and AI computing. Their core asset isn't ASICs or GPUs—it's the power contracts. Long-term, low-cost electricity agreements locked in during the mining boom. That's a structural advantage. Every AI data center operator needs cheap power. Core Scientific has it.
The rejected $9 billion acquisition—likely from a larger tech or infrastructure player—suggests the board and shareholders see a higher ceiling. The AMD partnership is the leverage point. But the market is mistaking a supply chain announcement for a revenue model.
Core: The Technical Reality of the AMD Deal
Here's where my experience as a trader who has audited mining-to-AI conversions comes in. I've seen this movie before. A mining company announces a GPU partnership. The stock jumps. Then the execution gaps emerge.
First, the technical challenges. Converting a Bitcoin mining facility to host AI workloads isn't trivial. Bitcoin miners are ASIC-based, air-cooled, and operate on a simple power-to-hashrate model. AI data centers require liquid cooling, high-density racking, and low-latency networking—InfiniBand or RoCE. Most mining sites don't have the physical infrastructure for this. Retrofitting costs are significant.
Second, the AMD factor. AMD's Instinct GPUs are competitive on paper, but the software ecosystem—ROCm—lags Nvidia's CUDA. For AI workloads, especially inference and training, CUDA is the de facto standard. AMD has made progress, but enterprise AI teams are still hesitant to switch. Core Scientific's customers will demand Nvidia. If the AMD partnership is a hardware-only arrangement, it's a diversification hedge, not a revenue driver.
Third, the lack of data. The announcement discloses zero operative metrics. No megawatts deployed. No utilization rates. No customer commitments. Zero. This is a strategic announcement, not a technical one. The market is treating it as validation. I see it as a placeholder.
Contrarian: The Real Value Is in the Power, Not the GPU
Here's the contrarian angle. The market is fixated on the AMD partnership as a catalyst. But the real value creation mechanism is the power infrastructure. Core Scientific's long-term PPA (Power Purchase Agreement) portfolio is the asset. The GPU is a commodity. The power is the moat.
Think about it. Any AI data center can buy AMD or Nvidia chips. But not every facility has access to 100MW of sub-4 cent per kWh power. That's the competitive edge. The AMD partnership is a signal that Core Scientific is serious about diversifying hardware supply, but it doesn't change the fundamental capital structure.
Shareholders rejected a $9 billion exit. That sets an implicit valuation benchmark. To justify that, Core Scientific needs to deploy AI capacity at scale. The AMD deal alone doesn't achieve that. They need to execute on the engineering, the customer onboarding, and the operational uptime.
Also, consider the debt load. Core Scientific emerged from bankruptcy with restructured debt. AI infrastructure requires massive capex. If they raise capital through equity dilution, the stock price could suffer. The AMD partnership might be a prelude to a capital raise, not a revenue contract.
Takeaway: Watch the MW, Not the Headlines
Core Scientific is a bet on infrastructure execution, not partnership optics. The AMD deal is a step, but it's not a finish line. The real signal will be when they announce deployed megawatts of AI capacity. If they can deliver 100MW of operational AI data center space within 12 months, the $9 billion rejection will look smart. If not, the market will punish the hubris.
Data over drama. Calculate. Execute. Repeat.
Liquidity vanishes. Lessons remain.
