Last week, Core Scientific’s shareholders turned down a $9 billion acquisition offer. The headline screamed confidence. Then came the AMD partnership announcement—a strategic alliance that promises to transform the company’s Bitcoin mining infrastructure into an AI computing powerhouse. But when you read between the code, you find a story that’s less about vision and more about survival. This is not a pivot. It’s a narrative of alchemy, where the market is betting that old mining hardware can be transmuted into gold-standard AI chips. And the data, so far, is silent.
Let me ground this in context. Core Scientific, a Nasdaq-listed cryptomining giant (CORZ), emerged from bankruptcy in early 2023 after the crypto winter wiped out many leveraged miners. Its core asset: owned and operated data centers filled with ASICs, tied to long-term power purchase agreements at below-market rates. The typical mining playbook—sell BTC to cover costs—was dying. The halving in April 2024 slashed block rewards by half, squeezing margins. The company’s survival strategy became clear: repurpose those same facilities for high-performance computing (HPC) and AI workloads. It’s a narrative that has captivated investors. The share price has tripled since the start of the year. But is the narrative supported by technical reality?
I’ve been tracking this “mining-to-AI” pivot since 2022, when I interviewed a dozen mining operators in Texas and Norway. My research showed that while the electricity and real estate are transferable, the compute stack is not. Bitcoin mining is embarrassingly parallel—ASICs do one thing, hash. AI training requires dense, low-latency interconnects, liquid cooling, and a software stack that can manage distributed GPU clusters. The conversion is not a plug-and-play. It’s a rebuild.
Now, the AMD partnership. The announcement was light on details—no MW commitment, no GPU count, no revenue share. The stock jumped 12% on the news. The market interpreted it as a validation of the pivot narrative. But here’s the unglamorous truth: AMD is desperate for data center deployments. Its Instinct MI300X chips are powerful, but the ROCm software ecosystem lags behind NVIDIA’s CUDA by several years. For a company like Core Scientific, adopting AMD means building a software stack from scratch, debugging compatibility issues, and convincing hyperscalers to trust a mixed-AMD deployment. The risk is not zero. The return is uncertain.
The $9 billion rejection sets a valuation anchor. Shareholders are saying, “We believe the company is worth more.” But a valuation anchor without a revenue bridge is just a wish. The AMD deal, if it materializes, could provide that bridge—but only if it includes binding power commitments and minimum GPU purchases. The article mentioned no such terms. Unearthing value where others see only chaos requires looking at the power contracts, not the press releases.
Let me break down the core narrative mechanics. The market is pricing in a future where Core Scientific captures a slice of the AI data center market, which is projected to grow to $300 billion by 2027. The narrative velocity is high: every miner that announces an AI pivot sees its stock jump. But the sentiment is fragile. If the company fails to deliver a single MW of AI capacity by Q3 2025, the narrative collapses. I’ve seen this pattern before—in 2020 with DeFi tokens that promised “cross-chain interoperability” but delivered only whitepapers. The narrative precedes the substance, and when the substance fails, the narrative exits first.
Now, the contrarian angle. What if the AMD partnership is a signal of weakness, not strength? Consider: Core Scientific’s largest competitor in the AI hosting space, CoreWeave, has exclusive contracts with NVIDIA for H100 and B200 GPUs. Core Scientific, by contrast, is turning to AMD—a second-tier supplier. This suggests they couldn’t secure adequate NVIDIA supply, or they’re taking a gamble on AMD’s future roadmap. In either case, the partnership may be a hedge, not a home run. The true blind spot is that the market assumes AMD’s software will catch up quickly. But from my experience auditing compute clusters, software maturity takes years, not quarters. Reading between the code to find the human story reveals engineers at AMD and Core Scientific working overtime to make ROCm stable, while the clock ticks on the Q4 2025 delivery date.
Another overlooked factor: the debt overhang. Core Scientific’s restructuring left it with significant debt, and the AI pivot requires massive capex—think $100 million per 100 MW of GPU capacity. The company may need to issue equity or take on more debt, diluting existing shareholders. The $9 billion valuation anchor assumes no dilution, but that’s naive. The AMD deal could include a debt component, further straining the balance sheet.
Let’s talk about the technical bottlenecks. Even if the software works, the physical infrastructure conversion is a multi-year process. Bitcoin mining facilities use air cooling and low-density racks (10-15 kW per rack). AI data centers require liquid cooling and high-density racks (40-60 kW per rack). The power distribution, backup generators, and network switches all need to be upgraded. Core Scientific has not disclosed how many of its facilities are suitable for conversion. My estimate, based on industry benchmarks, is that less than 30% of existing mining sites can be economically retrofitted. The rest are in remote locations with poor latency to major cloud on-ramps.
So where does this leave us? The narrative is strong, but the data is thin. The article’s core claim—that the AMD partnership is a strategic move to diversify chip supply—is true on the surface. But the deeper truth is that Core Scientific is betting its future on a complex, unproven conversion. The shareholders rejected a near-term exit, demanding long-term value. They may be right. But they may also be overestimating the company’s ability to execute. The narrative of value creation is only as strong as the technical milestones backing it.
My takeaway: Watch the Q3 2025 earnings call. If Core Scientific reports its first AI revenue from AMD-powered clusters, the narrative is validated. If not, the stock will correct sharply. The narrative velocity will slow, and the market will realize that infrastructure alchemy is harder than it looks. I’m not betting against the company—I’m betting on the data. And the data so far is a blank page.