Why Nvidia’s RTX Pro 6000 Blackwell Spiked 55% to $13,250 in a Year

In March 2025, Nvidia launched its RTX Pro 6000 Blackwell workstation GPU at an already-premium $8,565. Fast forward to June 13, 2026, and the official Nvidia Marketplace now lists the exact same card for $13,250 — a stunning 55% increase . It’s not a new model or a refreshed spec sheet. It’s a direct price hike driven by a simple but relentless equation: massive AI demand colliding with a severe VRAM shortage.

The card’s core specifications have not changed throughout this market drama. Built on the GB202 die, the RTX Pro 6000 Blackwell packs 24,064 CUDA cores, 752 fifth-generation Tensor Cores, and 96 GB of GDDR7 ECC memory on a 512-bit bus, delivering up to 1,792 GB/s of memory bandwidth . It draws 600 W through a PCIe 5.0 x16 interface and provides four DisplayPort 2.1b outputs . This makes it the most capable single-slot workstation GPU on the market for local AI workloads.

The Price Timeline in Detail

The card’s journey from premium to eye-watering unfolded in stages.

• March 2025 — Launch: The official MSRP was set at $8,565, with early US preorder listings around $8,435—$8,565 .
• Late 2025 — A Brief Dip: Preorder pricing briefly fell as low as $7,673 as early supply entered the channel .
• Early 2026 — Staying Below $8,000: For a window, several US retailers still offered the card for less than $8,000 . Amazon price trackers recorded an average around $9,072, with a low of $7,949 .
• May 2026 — The $10,000 Barrier Falls: Retail listings surged past $10,000–$11,000 in the US as AI demand intensified . In Central Europe, prices spiked from roughly €7,500 to €11,500 in a matter of days .
• June 13, 2026 — The Official Hike: Nvidia updated its own marketplace listing to $13,250, making the 55% increase official. The card is marked "Out of Stock" on Nvidia's store .

The Three Forces Inflating the Price

A combination of supply, demand, and a total lack of competition is responsible for the unprecedented price climb.

Global GDDR7 Shortage. The most significant supply-side constraint is a worldwide shortage of GDDR7 memory chips . Several Chinese-language reports identified the VRAM crunch as the single biggest hardware bottleneck, directly capping how many RTX Pro 6000 cards Nvidia and its partners can manufacture .

Unmatched AI Inference Demand. The RTX Pro 6000 is currently the only workstation GPU under $10,000—or at least, it was—that can fit a 70-billion-parameter model like Llama-3 70B entirely in a single card’s VRAM at Q4 quantization or higher . For enterprises, AI researchers, and edge deployments, this capability is not a luxury. It eliminates the complexity, latency, and cost of splitting large models across multiple GPUs. As AI inference workloads have exploded, so has the number of buyers willing to pay almost any price for that capability.

No Competitive Alternative. No other manufacturer offers a 96 GB workstation card in a single PCIe slot. This near-monopoly on high-VRAM professional GPUs gives Nvidia extraordinary pricing power . Without the pressure of a rival product, the market has absorbed price increases that would be unthinkable in a competitive segment.

Max-Q and Server Edition Pricing Also Rising

The RTX Pro 6000 Blackwell Max-Q Edition launched with the same $8,565 MSRP and identical core count but a halved 300 W TDP . By June 2026, a major US retailer listed the Max-Q at $10,999.99 . Similarly, eBay tracking data for the Server Edition in June 2026 shows an average lowest price around $12,498, with listings starting at $12,825 . The entire RTX Pro 6000 family has been pulled upward by the same market currents.

What This Means for Cloud GPU Rentals

For teams that cannot justify a $13,000+ capital expense, cloud instances remain the primary access point—and marketplace competition has kept rental rates far more stable than purchase prices.

Current cloud pricing for the RTX Pro 6000 as of June 2026:

• Floor pricing: $0.80/hr (Vast.ai, Nova Cloud on-demand single GPU)
• Median on-demand rate: ~$1.33/hr across 104 tracked configurations
• Spot/preemptible instances: ~$1.18—$1.19/hr (RunPod, Verda in US-East and EU-Central)
• Traditional on-demand providers: $1.71—$1.80/hr
• Dedicated GPU clusters: $1.80—$3.20+/hr depending on contract length and provider

Cloud pricing is expected to rise more slowly than retail because providers compete aggressively on per-hour rates. But sustained VRAM shortages could eventually feed through to higher rental costs if new hardware supply remains tight.

Will Prices Keep Climbing?

Several signals point to continued upward pressure, not a peak.

Nvidia’s own marketplace says “Out of Stock,” which makes the $13,250 MSRP somewhat theoretical; real-world street prices on eBay and secondary markets have already hit $12,825 to over $13,000 for the Server Edition . In Central Europe, user reports indicate week-over-week jumps of roughly €4,000 for the Workstation Edition . Chinese reseller data shows the card crossing $10,000 in May 2026 and then jumping another 32% month-over-month to reach $13,250 in June . As long as GDDR7 supply remains constrained and no competing 96 GB workstation card arrives, both the official MSRP and actual street prices are likely to move higher.

The RTX Pro 6000 Blackwell is a case study in how a specialized product with no rivals can become a pricing supernova when it sits at the intersection of a hardware shortage and an AI gold rush. For anyone who needs 96 GB of VRAM in a single slot today, there is one card, one supplier, and a very expensive bill.