While the market cheered Core Scientific's AMD partnership as a strategic pivot, a forensic examination of the data reveals a different story. Shareholders just rejected a $9B buyout, effectively betting that the company's new AI infrastructure play is worth more. But the on-chain volume—or lack thereof—of actual technical deliverables says otherwise.
Context: From Bankruptcy to AI Hype
Core Scientific emerged from Chapter 11 in early 2024, carrying legacy debt and a capital-intensive pivot from pure Bitcoin mining to hybrid HPC/AI colocation. The company now operates a fleet of Bitcoin miners and a growing portfolio of GPU clusters for AI workloads. The recent news: shareholders voted down a $9B acquisition offer, simultaneously announcing a partnership with AMD to deploy Instinct GPUs. The narrative being sold is that this partnership unlocks a new revenue stream worth more than the rejected bid.
Core: The Data That's Missing
Forensic mode: Activated. Let's strip away the marketing. The AMD press release mentions a 'multi-year collaboration' but provides zero granularity on:
- Committed megawatts (MW): How much power capacity has been allocated to AMD GPUs? Without this, we cannot size the revenue potential.
- Deployed GPU count: No numbers on Instinct MI300X or future chips.
- Utilization rates: For AI colocation, idle GPUs are a liability. Core Scientific hasn't disclosed how much of its existing infrastructure is already running at capacity.
- Software stack maturity: AMD's ROCm ecosystem still lags behind Nvidia's CUDA in developer tooling and library compatibility. Any AI workload migration requires engineering overhead—risk that is not priced in yet.
From my 2022 Terra crash forensics, I learned to trust transaction flows, not press releases. Here, the only verifiable 'transaction' is the shareholder vote rejecting $9B. That sets a floor valuation, but it does not prove the AMD deal will generate more value. The market is currently pricing in an optimistic scenario, but the data chain is broken: no delivery milestones, no SLAs, no revenue guarantees.
Contrarian: Correlation ≠ Causation
The popular interpretation is that the AMD partnership triggered the rejection, implying shareholders see more value in the standalone strategy. But correlation does not equal causation. The rejection could be driven by a belief that a better offer will emerge, or by hedge fund activism expecting a higher bid. The AMD deal itself may be a distraction—a way to boost the stock price before a future sale. The lack of disclosed terms suggests the partnership is more of a procurement agreement than a revenue-sharing joint venture. Without a minimum purchase commitment, AMD can walk away if its own GPU roadmap slips.
Takeaway: The Next-Week Signal
Over the next 30 days, watch for one number: the utilization rate of Core Scientific's existing data center capacity. If the company can show that its AI colocation is running at 70%+ with confirmed bookings, the AMD narrative gains credibility. If not, the stock may correct as the hype fades. Follow the power, not the press release.
