The ledger never lies, only the narrative obscures.
When a publicly traded Bitcoin miner refuses a $9 billion buyout and simultaneously announces a partnership with AMD, the market hears a story of defiant confidence. I hear the sound of missing data.
Let me be clear: I am not here to praise or bury Core Scientific (CORZ). I am here to read the on-chain and off-chain evidence—and in this case, the evidence is conspicuously absent. The company’s shareholder vote to reject the CoreWeave acquisition, paired with the AMD partnership announcement, creates a classic “narrative versus reality” divergence. My job, as a data detective, is to expose the gap.
Context: The Phoenix That Forgot to Burn
Core Scientific emerged from Chapter 11 bankruptcy in early 2024, carrying a legacy of debt and a pivot strategy: convert its Bitcoin mining infrastructure into high-performance computing (HPC) and AI data centers. The company signed multi-year hosting contracts with CoreWeave in 2024, signaling a credible move into AI cloud services. Then came the $9 billion offer from CoreWeave itself—a buyout that would have given shareholders an immediate exit. The shareholders said no.
Simultaneously, Core Scientific announced a partnership with AMD to deploy its Instinct GPUs. The press release was light on specifics: no committed megawatts, no performance benchmarks, no revenue guarantees. Just a strategic alignment.
From my 2017 ICO audits, I learned to distinguish between announcements and delivery. A whitepaper can promise a consensus algorithm; a press release can promise a GPU cluster. Neither is a substitute for verifiable data.
Core: The Evidence Chain Is Broken
Let’s examine what we actually know versus what we are supposed to believe.
Known fact #1: Shareholders rejected a $9 billion exit. That implies they believe the company is worth more. But “more” is not a number; it’s a hope. Known fact #2: AMD is supplying GPUs. AMD’s Instinct MI300X is a credible competitor to Nvidia’s H100, but its ROCm software ecosystem remains immature compared to CUDA. Migration costs, engineering overhead, and performance gaps are real.
Missing data points: - No disclosure of the contract value, duration, or minimum purchase commitments with AMD. - No technical validation of Core Scientific’s ability to retrofit its mining sites for AI workloads. Retrofitting requires liquid cooling, high-density racks, InfiniBand or RoCE networking, and GPU cluster orchestration—none of which are trivial for a company that previously managed ASICs. - No operational metrics: current AI capacity (MW), utilization rate, or customer uptake beyond the CoreWeave contract.
Correlation is a suggestion; causality is a truth. The market is correlating “AMD partnership” with “AI revenue growth.” But without a causal chain—specific engineering milestones, delivery schedules, and financial terms—this correlation is noise.
I built a Python script during the 2020 DeFi Summer to track APY sustainability across 12,000 liquidity pools. The lesson was simple: when you see high yield without a transparent breakdown of revenue sources, you are likely looking at a trap. Core Scientific’s AI pivot has a similar opacity. The headline is attractive; the underlying data is missing.
Contrarian: The AMD Deal Is a Hedge, Not a Home Run
Here is the counterintuitive angle: the AMD partnership may actually signal weakness, not strength.
Nvidia controls over 80% of the AI GPU market. Core Scientific’s existing CoreWeave contracts are presumably Nvidia-based. By adding AMD, Core Scientific is diversifying its supply chain—a prudent move. But it also reveals that the company cannot secure enough Nvidia GPUs at competitive prices to meet demand. AMD’s chips are less proven in large-scale production deployments, and the engineering costs of supporting two ecosystems (ROCm and CUDA) are non-trivial.
Furthermore, the shareholder rejection of the $9 billion buyout sets an implicit valuation floor. But valuation floors are only meaningful if the company can generate cash flows to justify them. Core Scientific’s mining revenue is tied to Bitcoin’s price, which is volatile. Its AI hosting revenue is under long-term contracts, but those contracts are with CoreWeave—the same company that tried to buy it. The conflict of interest is palpable.
Trust the hash, not the headline. The hash rate of a Bitcoin mine is transparent; the hash rate of an AI data center is not. Until Core Scientific publishes audited operational data—delivered megawatts, utilization rates, customer churn—the AMD partnership remains a narrative device, not a technical milestone.
Takeaway: Watch the Watts, Not the Words
In the next 90 days, I will be watching for two signals: (1) a public disclosure of AMD GPU deployment numbers and associated power draw, and (2) any update on the CoreWeave contract renegotiation. If the company cannot provide these, the $9 billion rejection will look like a mistake, not a conviction.
An algorithm does not sleep, nor does it feel fear. Neither should investors.