You think the rejection of a $9 billion acquisition is a vote of confidence? Logic doesn't care about press releases. On February 12, 2025, Core Scientific shareholders officially turned down a $9 billion buyout offer. The same day, the company announced a partnership with AMD. The market cheered. I see a different story: a structural bet on a conversion that hasn't been proven, wrapped in a strategic partnership that lacks technical depth.
## Context: The Infrastructure Pivot Core Scientific is a publicly traded company (CORZ) that operates Bitcoin mining facilities. Post-bankruptcy in 2023, it has been repositioning itself as an AI data center host. The rejected offer—reportedly from a consortium—would have taken the company private. The AMD partnership is the latest move to diversify chip supply beyond Nvidia, aiming to retrofit mining sites for GPU-based AI workloads. This is not a protocol upgrade; it's a physical infrastructure transformation.
## Core: The Dissection ### Technical Appraisal: The Partnership is a Token, Not a Promise Let’s start with the AMD deal. The announcement mentions no technical details: no delivered megawatts, no GPU cluster size, no network topology, no software stack validation. Based on my experience auditing infrastructure transitions, a partnership is a handshake, not a deployment. AMD needs real data centers to test its Instinct GPUs against Nvidia’s dominance. Core Scientific needs GPU supply. This is a mutual dependency, not a solved problem.
The real technical challenge is the retrofit. Converting a Bitcoin mine to an AI data center requires liquid cooling, high-density racks, InfiniBand or RoCE networking, and GPU cluster scheduling. Bitcoin mining is a parallel, low-latency operation. AI training is a distributed, high-bandwidth serial computation. The two are not the same. AMD’s ROCm software stack still lags CUDA in maturity. I don’t trust a press release to solve that.
### Structural Incentive Dissection: The Shareholder Vote as a Valuation Anchor Rejecting a $9 billion offer sets a floor. The market now expects management to create at least $9 billion in equity value. This is a high bar for a company that emerged from bankruptcy with debt and a capital-intensive pivot. The article’s framing that the AMD partnership “outweighs” the rejected deal is not a fact—it’s a narrative. Greed is the feature; the bug is just the trigger. Shareholders are betting on a future that requires massive capital expenditure. Expect dilution via stock offerings or convertible debt.
You didn’t buy a token; you bought a stock. No native token means no burn mechanisms, no staking yields. Value capture is purely through earnings and multiple expansion. The revenue split for the AMD deal is undisclosed. If it’s a simple procurement agreement, Core Scientific bears the risk of GPU utilization. If it’s a hosting deal, the revenue is more predictable but still depends on AMD’s ability to deliver chips on time. The exploit wasn’t a code bug; it was a strategic pivot unsupported by technical milestones.
### Mathematical Rigor: The Missing Metrics A healthy AI infrastructure company reports: deployed MW, utilization rate, power purchase agreement terms, and client count. This article mentions none. Core Scientific’s pivot is engineering, not magic. The power cost advantage from long-term mining contracts is a real asset. But converting that into a competitive AI hosting offering requires capital, time, and operational excellence. The AMD partnership does not address the network bottleneck: the ability to interconnect thousands of GPUs for training. Without that, it’s just a GPU farm.
## Contrarian: What the Bulls Got Right Let’s be fair. The conversion of mining infrastructure to AI data centers is technically feasible. Core Scientific has physical assets, power contracts, and existing relationships. AMD’s Instinct GPUs are a credible alternative to Nvidia for specific workloads. The shareholders’ rejection of a $9 billion offer signals confidence in the company’s long-term value. The energy infrastructure is a real barrier to entry—most AI cloud providers don’t have 100+ MW of shovels-ready sites.
The contrarian view is that the market is underpricing the optionality. If Core Scientific successfully executes, the AI hosting revenue could dwarf its mining revenue. The partnership with AMD also reduces supply chain risk. The bulls argue that the market sees a transformation, not a turnaround. I acknowledge that the thesis is not irrational—it’s just unverified.
## Takeaway: The Accountability Call Core Scientific’s shareholders have drawn a line in the sand. Now the company must deliver. The AMD partnership is a potential catalyst, but it’s also a liability. The next quarterly report will need to show MW under contract, not just partnership announcements. The market will treat the $9 billion rejection as a floor—and every missed milestone will be a call to sell. The real test is not the partnership; it’s the execution. You didn’t buy a token; you bought a bet on a conversion that hasn’t happened yet. Logic doesn’t care about your narrative. It only cares about delivered megawatts.