The market is buzzing about Core Scientific’s partnership with AMD. Headlines scream ‘strategic pivot to AI,’ and the stock ticker CORZ is catching institutional attention. But ignore the headlines. Watch the flow. The real story is not the chip deal—it’s the shareholder revolt that rejected a $9 billion acquisition. That vote is a liquidity signal, and it’s telling us something the AMD press release will never reveal.
I’ve seen this pattern before. In 2017, I watched ICO teams parade partnerships with Microsoft and Amazon Web Services, while their tokenomics were bleeding value. The gloss of a name-brand collaboration masks the fundamental question: where is the cash flow coming from, and who is paying for it? Core Scientific’s shareholders just answered that question with a resounding ‘no.’ They don’t believe the company is worth $9 billion today—they believe it’s worth more, but only if management can deliver on a transformation that is still unproven.
Context: From Bankruptcy to AI Hype
Core Scientific is not a fresh startup. It’s a Nasdaq-listed Bitcoin miner that emerged from Chapter 11 bankruptcy in early 2024. The company operates massive power infrastructure—think warehouses filled with ASICs for Bitcoin mining. But the mining business has a razor-thin margin post-halving, and the entire industry is scrambling to repurpose those power assets for AI compute. The idea is elegant: take the cheap electricity, cooling systems, and real estate that were optimized for SHA-256 hashing, and install GPUs for AI inference and training. Core Scientific signed a multi-year hosting contract with CoreWeave, a pure-play AI cloud provider, in 2024. That gave them credibility. Then came the AMD announcement.
AMD’s Instinct GPUs are positioned as a competitor to Nvidia’s H100/B200 line. The partnership suggests Core Scientific will deploy AMD hardware in its data centers. But the details are thin. No capacity commitments, no revenue guarantees, no timeline. Compare that to the $9 billion acquisition offer that was rejected. That offer was a concrete number. The AMD partnership is a statement of intent, not a contract. The market is treating them as equivalent, but that is a mistake.
Core: The Technical Reality of the AMD Deal
Let’s break down what this partnership actually requires. Core Scientific needs to retrofit its Bitcoin mining facilities to support AI workloads. That means installing liquid cooling, InfiniBand or RoCE networking, high-density power distribution, and GPU cluster management software. These are not off-the-shelf solutions. The company has experience with ASICs, but GPUs are a different beast. ASICs are fixed-function—they mine Bitcoin, period. GPUs are programmable, but they require a software stack like CUDA (Nvidia) or ROCm (AMD). ROCm is open-source and improving, but it lags behind CUDA in maturity, developer tooling, and library support. Any AI workload that relies on Nvidia’s cuDNN or TensorRT will need to be ported, and that’s a friction not every customer will accept.
Core insight: The AMD partnership is a supply-chain diversification play, not a technological leap. Core Scientific is hedging against Nvidia’s dominance, but AMD’s hardware is still unproven at scale for the same workloads. I’ve audited infrastructure projects that claimed to be ‘AI-ready’—the reality is that power delivery and cooling are the bottlenecks, not the chip brand. The company’s real asset is its long-term power purchase agreements (PPAs) locked in at low rates. That’s the economic moat, not the GPU sticker.
From a quantitative perspective, the article lacks any operational metrics. How many megawatts of AI capacity has Core Scientific actually delivered? What is the utilization rate of its CoreWeave contract? What is the margin on AI hosting versus mining? Without these numbers, the AMD partnership is a narrative, not a valuation. I’ve seen this dynamic in DeFi—projects announce a liquidity mining program, the token pumps, but the underlying yield is subsidized by inflation. DeFi yields are traps, not gifts. The same logic applies here: partnership announcements are traps for investors who confuse press releases with revenue.
Contrarian: The Decoupling Thesis
The conventional wisdom is that AMD partnership equals bullish for Core Scientific. I see the opposite. The shareholder rejection of the $9 billion sale is a vote of no confidence in management’s ability to execute the AI pivot. Think about it: a private buyer was willing to pay $9 billion for the entire company. That buyer saw value in the power infrastructure, the mining fleet, and the potential AI revenue. But shareholders said ‘no’—they want the stock to be worth more under current management. That’s a bet on execution, not just on the macro trend.
Here’s the contrarian angle: The AMD partnership might actually be a distraction. Core Scientific is still a miner first. Bitcoin mining revenue is volatile, and the company’s balance sheet is still recovering from bankruptcy. AI infrastructure requires massive capital expenditure—building out GPU clusters costs billions. Where will the money come from? The company could issue new shares, diluting existing holders. Or it could take on debt, increasing leverage. The $9 billion offer would have provided a clean exit. Now, shareholders are exposed to the risk of a capital-intensive transformation with no guarantee of returns.
Watch the flow, ignore the noise. The liquidity trail points to a cash crunch. The company’s stock price has rallied on the AMD news, but that’s speculative—it’s pricing in perfection. If the AI pivot stumbles—say, ROCm compatibility issues or slower-than-expected capacity deployment—the stock will correct. The shareholder vote created a floor, but it also raised the bar. Management now has to deliver value above $9 billion. That’s a high hurdle.
I’ve been in this position before. In 2021, I advised a fund to short exposure to NFT marketplaces that were trading at absurd P/E ratios. The market was euphoric, but the fundamentals were broken. NFTs are digital vanity metrics. The same applies here: the AMD partnership is a vanity metric for Core Scientific. It looks good on paper, but it doesn’t change the underlying economics of power conversion and compute utilization.
Takeaway: Positioning for the Next Cycle
The question is not whether Core Scientific will succeed in AI, but whether the current stock price already reflects that success. The shareholder vote tells us that the board believes the company is worth more than $9 billion. That’s a signal, but it’s not a guarantee. The AMD partnership adds optionality, but it also adds execution risk. For a fund manager like me, the play is to wait for actual capacity numbers. When the company announces that it has deployed 100 MW of GPU compute, with a customer list and utilization rates, then I’ll pay attention. Until then, this is a narrative trade.
Arbitrage closes; liquidity remains. The arbitrage between the $9 billion offer and the current market cap is closing, but the underlying liquidity of the stock and the business remains uncertain. Core Scientific is a bet on the convergence of Bitcoin mining and AI—a thesis I believe in, but not at any price. The next 12 months will reveal whether the company can execute. I’ll be watching the power meters, not the press releases.
