Shareholders just told Core Scientific’s board they’d rather bet on Bitcoin mining rigs and AMD GPUs than a $9 billion check. That’s a bold move for a company that crawled out of bankruptcy less than two years ago. The AMD partnership? It’s a headline, not a technical roadmap. t check.
Let’s rewind. Core Scientific is a Bitcoin miner—one of the big ones. They operate massive facilities running ASICs, sucking up cheap power secured through long-term contracts. That’s their moat. But after the 2022 crypto winter and the 2023 bankruptcy, they’ve been pivoting. The story? Convert those power-rich mining sites into AI data centers. Rent out GPU compute to hungry AI startups. It’s the same narrative that’s been pumping miner stocks all year: Bitcoin miners as AI infrastructure plays.
In late 2024, they landed a deal with CoreWeave, a pure AI cloud provider. Then came the AMD announcement. And then, the kicker: an anonymous suitor offered $9 billion to buy the whole company. Shareholders said no. They’re banking on the AMD partnership to deliver more value than that offer.
Pump, dump, debug. Repeat.
Here’s what we actually know. The AMD partnership exists. That’s it. No financial terms, no minimum purchase commitments, no delivery schedule, no performance benchmarks. The press release says “collaboration” but doesn’t say how many Instinct GPUs are being deployed, what workloads they’ll run, or when the first rack goes live. For a company that emerged from Chapter 11, that’s a thin thread to hang a $9 billion valuation on.
I’ve been tracking this pivot for months. From my days debugging Solidity and auditing mining pool contracts, I’ve learned to separate operational reality from narrative. The conversion from ASIC mining to GPU AI hosting is not a plug-and-play upgrade. It’s a complete infrastructure overhaul. ASICs are custom chips for SHA-256. GPUs are general-purpose compute, but they need dense networking (InfiniBand or RoCE), high-power cooling systems (liquid cooling for H100s, still experimental for Instinct), and completely different power distribution. A mining facility running 100 MW can handle ASICs at 30-40 MW per building. For GPUs, the same power envelope might support only a fraction of the racks, because GPU clusters are power-hungry and heat-sensitive.
And then there’s the software stack. AMD’s ROCm ecosystem is maturing, but it’s not CUDA. Most AI training frameworks are optimized for Nvidia. If Core Scientific deploys Instinct GPUs, their customers will need to port their models. That’s friction. That’s cost. That’s time. The AMD partnership might be a strategic hedge, but it’s not a silver bullet.
Gas fees higher than the yield. Typical.
Let’s talk about the rejected $9 billion. That’s not a random number. It’s a valuation anchor. Shareholders are effectively saying: “We believe the company is worth more than $9 billion in the long run.” But that belief is based on future AI revenue, not current operations. As of now, Core Scientific reports mining revenue and some hosting income. Their AI hosting revenue is not yet material. The market is pricing in a successful pivot, but the technical proof is missing.
From my hands-on experience testing AI agent economies in 2026, I know that deploying autonomous agents on stablecoins was a nightmare of friction points. The same lesson applies here: infrastructure pivots are harder than they look. The hype cycle runs ahead of the engineering cycle. The AMD partnership might not deliver tangible capacity for 18 months. By then, the bull market could cool, and the cost of capital could rise.
Here’s the contrarian angle: The real value of Core Scientific isn’t the AMD partnership. It’s the power contracts. Those long-term, fixed-price electricity agreements are gold in a world of volatile energy costs. But converting that power into AI compute is a capital-intensive transformation. The company needs to raise billions to retrofit sites, buy GPUs, and build networking. The AMD partnership might just be a way to secure GPU supply without paying upfront—a vendor financing deal. But if AMD’s chips don’t perform, the company is stuck with a portfolio of underutilized hardware.
And the market is already euphoric. Miner stocks are flying. Core Scientific’s stock price has doubled since the AMD announcement. Green candles blind people to red flags. The shareholders who rejected the $9 billion might be suffering from the same market-induced optimism. They’re betting on a perfect execution path with no margin for error.
t check.
Look at the competitors. CoreWeave, the AI cloud provider, has real GPU clusters, live contracts, and documented performance. They’re not converting mining sites; they’re building from scratch. Core Scientific is trying to do both: maintain mining operations while building a new AI business. That’s a dual focus that often leads to mid execution.
What’s the next watch? Forget the press releases. The only metric that matters is megawatts delivered to AI customers. How many racks are live? What’s the utilization rate? What’s the average revenue per megawatt? If Core Scientific doesn’t report these numbers by Q3, the market will start asking hard questions. The AMD partnership is a narrative booster, but it’s not a technical milestone.
I’m watching the next earnings call, not the next partnership announcement. If they can show operational progress—like 50 MW of GPU capacity online with committed customers—then the $9 billion rejection might look smart. If not, it’ll be another case of the market falling for a story without code.
Pump, dump, debug. Repeat.
