The demand for raw compute power has gone off the charts, and TinyCorp, an AI startup that lives on token‑based economics, has set its sights on a monster‑sized RDNA 5 graphics card from AMD. The company’s pitch to investors hinges on raising roughly $11.5 million to power a 5 MW data center in Oregon, where it plans to outfit 3,000 custom rigs with next‑generation AMD silicon. If the cards materialize, TinyCorp expects each unit to generate a modest slice of revenue by selling compute time on platforms like OpenRouter, ultimately targeting a $5.4 million cash flow from the token‑sale model.
AMD’s roadmap does indicate a RDNA 5 launch around mid‑2027, but the notion of a consumer‑grade GPU packing 96 GB of VRAM is, at best, speculative. History shows that AMD’s RX 9000 series struggled with VRAM limits, and the global DRAM shortage only makes a 96‑GB configuration look impossible for any mainstream product. The only real precedent for that memory size today is NVIDIA’s RTX PRO 6000 Ada Blackwell, which retails for $8,000‑$10,000, a price point that dwarfs TinyCorp’s projected $2,500 per card.
In reality, expecting AMD to ship a 96‑GB RDNA 5 card for $2,500 is a pipe‑dream that could stall the entire venture. The only plausible path for such a massive memory pool would be through AMD’s workstation‑focused Radeon PRO line, and even that remains unconfirmed. TinyCorp even hinted it would design its own board around the chip if AMD fails to deliver a comparable model, a move that mirrors the DIY spirit seen in the Blackwell community.
For a tech‑savvy audience that’s watched GPU shortages turn consumer cards into AI workhorses, the lesson is clear: while ambition fuels innovation, grounding expectations in the realities of silicon production and memory supply is essential. TinyCorp’s vision is certainly bold, but without a concrete AMD roadmap, the dream of a 96‑GB consumer RDNA 5 GPU may remain just that, a dream.
Gladstone is a tech virtuoso, boasting a dynamic 25-year journey through the digital landscape. A maestro of code, he has engineered cutting-edge software, orchestrated high-performing teams, and masterminded robust system architectures. His experience covers large-scale systems, as well as the intricacies of embedded systems and microcontrollers. A proud alumnus of a prestigious British institution, he wields a computer-science-related honours degree.