The market digested the news with a collective shrug. Core Scientific shareholders rejected a $9 billion acquisition offer, and the stock barely flinched. Then came the AMD partnership announcement, and the narrative shifted from 'what did they reject' to 'what did they gain.' As a trader who has watched infrastructure narratives rise and fall since the 2017 ICO boom, I see something deeper: a bet that physical power contracts are worth more than any buyout premium. The ledger remembers what the market forgets: value is built in kilowatts, not press releases.
Context: From Bankruptcy to Bifurcation
Core Scientific is not a new player. It emerged from Chapter 11 in early 2023, its balance sheet scarred by the 2022 crash. The company’s core asset is not GPUs or ASICs—it is a portfolio of long-term power purchase agreements (PPAs) locked in at low rates, often from stranded energy assets. This is the same infrastructure playbook that made Bitcoin mining a viable business: take cheap, otherwise wasted energy, and convert it into computational output. Now, they are attempting to remix that formula for AI.
The company operates two distinct revenue streams: Bitcoin mining (using ASICs) and AI hosting (using GPUs and specialized hardware). The mining side is a commodity business—revenue is a function of hash rate, bitcoin price, and network difficulty. Post-halving, the unit economics have tightened; miners with the cheapest power survive. The AI side, however, is a services business with long-term contracts and higher margins. Core Scientific signed a multi-year hosting deal with CoreWeave in 2024, committing to deliver 200 megawatts of AI-ready infrastructure. The AMD partnership is an extension of this pivot: access to AMD Instinct GPUs to service a growing demand for alternatives to Nvidia.
But the rejection of the $9 billion buyout—reportedly from a private equity consortium—signals something else. Management believes the company is worth more than that. The AMD partnership is the proof point they are offering. The question is whether the market will agree.
Core: The Infrastructure Gap and the GPU Supply Chain
Let me pull from my own technical experience. In 2017, I audited smart contracts for a syndicate in Ho Chi Minh City. I watched a project called VictoryCoin lose $400,000 to a simple integer overflow. The code was sound in theory, but the human layer—the assumptions about how it would be used—was flawed. That experience taught me that infrastructure is never just hardware; it is the operational and cultural context in which it operates.
Core Scientific’s transformation from a mining operator to an AI hosting provider is not a simple hardware swap. It requires a complete overhaul of the data center architecture:
- Cooling: ASICs are air-cooled; GPUs for AI training require liquid cooling or advanced air handling at higher densities. Core Scientific has been retrofitting its existing facilities, but the cost and complexity are non-trivial.
- Networking: While Bitcoin mining only needs internet connectivity, AI workloads require high-bandwidth, low-latency interconnects (InfiniBand or RoCE). This is a different skill set and capital expenditure.
- Software Stack: The AMD Instinct GPUs are not drop-in replacements for Nvidia. The ROCm software ecosystem has matured, but it still lags CUDA in terms of library support, developer tooling, and community adoption. Core Scientific’s customers are likely to be AI startups or enterprises that are cost-sensitive or strategically avoiding Nvidia lock-in. But the operational burden of managing a non-CUDA stack is real.
The AMD partnership is more than a procurement deal. It is a strategic alignment. AMD needs reference deployments to prove its GPUs are viable for AI workloads. Core Scientific needs a reliable supply of GPUs to attract customers. Both parties are betting on a market that is currently dominated by Nvidia. The 2022 winter taught me that diversification is not just a hedge; it is a survival strategy. But the execution risk is immense.
Let me offer a concrete metric. The market should focus on megawatts delivered and utilization rates. Core Scientific has committed to 200 MW for CoreWeave. The AMD partnership will likely add another 50-100 MW over the next 18 months. If they can deliver on time and achieve utilization above 70%, the revenue profile will justify the rejection of the $9 billion offer. But if they miss these milestones, the stock will trade at a discount to the buyout price.
Contrarian: The Hype Cycle and the Forgotten Debt
The narrative around Core Scientific is overwhelmingly positive: the pivot to AI, the AMD partnership, the rejection of a lowball offer. But I see three blind spots that the market is ignoring.
First, the capital structure is fragile. The company emerged from bankruptcy with a clean balance sheet, but the AI transformation requires massive CapEx. The cost to retrofit a 100 MW facility from mining to AI hosting is estimated at $50-100 million. Core Scientific will need to raise capital—either through equity dilution (which the shareholder vote suggests they want to avoid) or through debt. Debt markets are open for AI infrastructure, but the terms are punitive. The company’s debt-to-equity ratio could rise sharply, increasing the risk profile.
Second, the AMD partnership is a two-edged sword. While it provides supply diversification, it also signals that Core Scientific could not secure enough Nvidia GPUs. The market is currently paying a premium for any AI infrastructure story, but when the AI hype cycle cools—and it will—the premium will disappear. At that point, the company will be left with a hybrid infrastructure that is neither the cheapest Bitcoin miner nor the most efficient AI cloud. It will be stuck in the middle.
Third, the shareholder rejection of $9 billion is a vote of confidence, but also a bet on execution. The management team has a track record of bankruptcy. They are not the same team that led the company into Chapter 11, but the institutional memory of failure is still present. The ETF approval in 2024 has brought institutional capital into the space, but that capital is unforgiving. If Core Scientific misses a guidance number, the stock will be punished severely.
I recall the 2020 DeFi summer. I watched friends chase 1000% APYs while I moved my capital into low-risk stablecoin pools. They laughed at my caution. Six months later, many of those protocols were dead. The contrarian move is not always the popular one. Here, the contrarian view is that the AMD partnership is a narrative crutch, not a fundamental transformation. The real value of Core Scientific is its power contracts, not its GPU partnerships.
Takeaway: The Next 12 Months
Over the next year, the market will separate signal from noise. The key metrics are not press releases; they are operational data:
- Megawatts under contract for AI hosting: Current is ~200 MW with CoreWeave. Any addition from AMD will be a positive signal.
- Utilization rate of AI infrastructure: Above 70% indicates strong demand; below 50% suggests overcapacity.
- Debt-to-equity ratio: If it rises above 1.5, the market will discount the equity.
- Bitcoin mining revenue: As a percentage of total revenue, this should decline. If it stays above 40%, the AI pivot is not real.
The rejection of $9 billion is a bold statement. It forces management to deliver. I will watch the next earnings call with the same intensity I once watched a smart contract audit. The code is not neutral; neither is the balance sheet. We traded souls for pixels, now we seek the ghost. The ghost is the trust that the company will execute. Between the block and the breath, truth resides. The truth will be in the kilowatt hours delivered, not the headlines.
As I write this, I am reminded of my time in the Mekong Delta in 2022, isolated from the noise, studying zero-knowledge proofs. That period taught me that the only way to survive the bear market is to focus on fundamentals. Core Scientific is a fundamental bet on the convergence of mining and AI. But the convergence is not guaranteed. The algorithm does not care about your conviction. The algorithm only cares about cash flow.
For the reader who is waiting for direction: the chop is for positioning. If you believe in the infrastructure convergence, accumulate on dips below the $9 billion implied valuation. But set a stop at 20% below that level. The market will reward patience, but it will punish hubris.
FOMO is the tax on unexamined desire. The desire to own a piece of the AI future is understandable. But examine the footnotes. The AMD partnership is a step, not a leap. The true test will come in 18 months, when the first AMD-powered cluster is fully operational. Until then, trade the narrative, but know that the narrative is a ghost. The flesh is in the power contracts.
The ledger remembers what the market forgets. I will remember that a $9 billion rejection is a bet on the future. But the future is never certain. Silence in the code screams louder than volume. Listen to the numbers.