Will the monthly average compute price of NVIDIA's RTX 5090 be above $2.00 in December 2026?

- [2026-09-22] A related Kalshi/Coinbase prediction market ("NVIDIA RTX 5090 · Hourly price on Oct 09, 2026") that resolves on the same Ornn index shows the "Above $0.50" strike trading around 78% Yes, "Above $0.55" around 48%, and "Above $0.60" around 50% — implying the market-implied current/near-term RTX 5090 Ornn compute price sits close to $0.50–$0.60/hr, nowhere near the $2.00 threshold. If the value of NVIDIA RTX 5090 compute per hour is above $0.6 on Oct 09, 2026 at 4 PM ET, then the market resolves to Yes, based on the value of NVIDIA RTX 5090 compute per hour as reported by Ornn. (https://www.coinbase.com/en-fr/predictions/event/KXRTX5090WS-26OCT09) - [2026-06-18] Kalshi (and mirrored on Coinbase) launched GPU compute-price prediction markets settling on Ornn's index; the resolution source is explicitly the Ornn compute index, not retail card prices. Prediction markets company Kalshi has added the ability to gamble on the compute per hour cost of Nvidia GPUs, verified based on data from Ornn AI, which tracks live traded spot prices across hardware including H100, H200, B200, and RTX 5090, with H100s currently at $1.70. (https://www.datacenterdynamics.com/en/news/kalshi-users-able-to-gamble-on-nvidia-gpu-compute-prices-based-on-ornns-compute-derivatives-platform/) - [2026-09-21] Ornn's public index dashboard (data.ornn.com) settles H100 SXM at $2.79/hr, H200 at $5.07/hr, B200 at $7.71/hr, and A100 SXM4 at $0.99/hr as of Sept 21, 2026 — showing that even the most expensive datacenter-grade GPUs on this index sit at or below ~$2.79-7.71/hr, and RTX 5090, a much cheaper consumer-class card, is tracked as a fifth GPU on the same index. (https://data.ornn.com/preview) - [2026-09-20] Independent GPU-rental aggregator GetDeploying reports: "As of September 20, 2026, the median on-demand price is $0.67 per GPU per hour across 16 providers with a priced on-demand config" for the RTX 5090. (https://getdeploying.com/gpus/nvidia-rtx-5090) - [2026-09-19] Aggregator gpus.io reports: "Currently available from 9 providers starting at $0.41/GPU/hour, with a market median of $0.80/GPU/hour across 70 configurations" for the RTX 5090. (https://gpus.io/en/gpus/rtx5090) - [2026-09-19] Aggregator GPU Finder notes: "The 1× RTX 5090 on-demand floor is $0.27/hr ($0.27/GPU-hr), up 33% vs last month — 6-month range $0.20–$0.65/hr for 1× nodes" , confirming the whole recent trading range has stayed well under $1/hr. (https://gpufinder.dev/gpu/rtx-5090) - [2026-09-17] Cross-provider index AIMultiple states current RTX 5090 median at roughly $0.63/hr: "RTX 4090 and RTX 5090 sit below a dollar, at $0.44 and $0.63 respectively" for September 2026. (https://aimultiple.com/gpu-index) - [2026-09-19] Runcrate lists RTX 5090 cloud rentals from $0.45/hr, and llmhosting.ai lists them from $0.268/hr, consistent with the sub-$1 range seen across nearly all trackers. (https://www.runcrate.ai/pricing/gpu/rtx-5090, https://llmhosting.ai/gpus/rtx-5090) - [2026-09-20] Retail (physical card) prices for the RTX 5090 have surged separately to $4,300–$7,000+ due to a GDDR7/DRAM shortage, but this retail-card price is a distinct metric from the hourly cloud-compute rental price that the Ornn index and this Kalshi market track; retail scarcity has not translated into datacenter/cloud rental rates above $1/hr. (https://tech-insider.org/gpu-prices-2026/) - [2026-03-31 to present] Kalshi's RTX 5090 monthly markets have existed since March 2026 with initial contracts covering RTX 5090 and H200 per-hour rates set to resolve by March 31, 2026 , and no source found indicates the RTX 5090 hourly rate has ever approached $2.00 in any month since launch. BOTTOM LINE: Every independent tracker (GetDeploying, gpus.io, GPU Finder, AIMultiple, Runcrate, llmhosting.ai) and the Ornn-linked prediction market itself put the RTX 5090 cloud compute price in the roughly $0.25–$0.85/hr range as of mid-to-late September 2026, with related Ornn-based markets pricing the odds of even a $0.60 threshold at only ~50% for early October — implying the December 2026 monthly average would need to roughly triple-to-quadruple current levels to clear $2.00, which no current data or trend supports. NOT FOUND: A direct historical monthly-average RTX 5090 value from the Ornn Compute Price Index (data.ornn.com) itself, since the RTX 5090 series was not shown in the free/preview data tier at the time of the fetch (only H100, H200, B200, A100 were listed); also could not find the actual current Kalshi "December 2026" RTX 5090 market page or its live trading odds.

- [2026-09-22] Multiple independent GPU-rental aggregators (CloudRift, GPU.ai, GetDeploying, GPUPerHour, GPU Finder) report current RTX 5090 on-demand compute rental rates clustering roughly $0.35–$1.00/GPU-hour, with a median around $0.67/hr; "As of September 20, 2026, the median on-demand price is $0.67 per GPU per hour across 16 providers with a priced on-demand config." https://getdeploying.com/gpus/nvidia-rtx-5090 - [2026-09-19] An independent tracker put the cheapest confirmed in-stock RTX 5090 rental at $0.27/GPU-hr, noting the "1× RTX 5090 on-demand floor is $0.27/hr ($0.27/GPU-hr), up 33% vs last month — 6-month range $0.20–$0.65/hr for 1× nodes" , indicating the whole 6-month trading window has stayed far below $2. https://gpufinder.dev/gpu/rtx-5090 - [2026-04-09] A prior Kalshi/Octagon-tracked RTX 5090 compute-price market had a strike of just $0.4708, showing the Ornn index (the same data source underlying this Kalshi question) was near $0.47/hr in April 2026 — a base level roughly 4x below the $2 threshold. https://octagonai.co/markets/science-and-technology/energy/nvidia-rtx-5090-compute-price-up-or-down-by-apr-10-2026/ - [2026-09-21] A currently-open weekly Coinbase/Kalshi-style RTX 5090 compute market for Oct 9, 2026 uses a strike of $0.60, confirming the Ornn index (dashboard.ornnai.com) is still trading in the sub-$1 range seven months into the year, not near $2. https://www.coinbase.com/predictions/event/KXRTX5090WS-26OCT09 - [2026-09-21] Ornn's free public index (the settlement source for these Kalshi contracts) tracks H100 ($2.79/hr), H200 ($5.07/hr), B200 ($7.71/hr) and A100 ($0.99/hr) — all far higher-value data-center-grade chips priced above $2/hr, unlike the RTX 5090 which is a consumer card with comparatively low compute-index pricing. https://data.ornn.com/preview - [2026-09-21] GetDeploying's broad GPU price-trend index shows overall cloud GPU on-demand rental prices are roughly flat over the trailing 4 weeks and up only about 9% over the trailing 12 months across 22 tracked models, arguing against a rapid, market-wide multi-fold rental-price spike by December. https://getdeploying.com/gpu-price-trends - [2026-09-01] Retail/street prices for physical RTX 5090 cards have risen sharply (roughly $4,300–$5,200, more than double MSRP) due to memory/AI-driven scarcity, but this retail price surge has not translated into a comparable surge in hourly rental/compute pricing, which remains driven by neocloud/spot-market competition rather than card scarcity alone. https://tech-insider.org/gpu-prices-2026/ - [2026-09-20] Reference class check: no historical Kalshi/Ornn-tracked RTX 5090 monthly or weekly compute-price market has been found where the index cleared $1/hr, let alone $2/hr — every observed strike level (weekly and monthly) since markets launched in March 2026 has been set well under $1 (e.g., $0.47 in April, $0.60 in October), consistent with an index that has drifted up only modestly across two-plus quarters. BOTTOM LINE: The reference class of RTX 5090 "compute per hour" (rental) pricing — as distinct from retail card prices, which have spiked to $4,000–$5,000+ — has stayed in a roughly $0.20–$1.00/hr band throughout 2026 and only crept up gradually (from ~$0.47/hr in April to ~$0.60–$0.67/hr in September), so historically and structurally this metric has never come close to the $2/hr threshold; a base-rate/outside-view read strongly favors "No." NOT FOUND: Could not locate the exact settled Ornn RTX 5090 index values for each individual month (Jan–Aug 2026) or a full historical Kalshi resolution log confirming how many prior monthly RTX 5090 markets settled Yes vs No; could not access full historical OCPI RTX 5090 chart data (subscription-gated) to quantify month-over-month trend precisely.

- [2026-03-XX] Kalshi launched GPU compute prediction markets settling to Ornn's index; at launch coverage noted Ornn reported H100 compute prices averaging around $1.70 per hour , establishing that GPU compute pricing on this index can sit well above $1/hr for premium data-center chips, giving some precedent for elevated compute prices generally. https://www.kucoin.com/news/flash/kalshi-launches-prediction-markets-for-gpu-compute-power-prices - [2026-04-09] A Kalshi RTX 5090 weekly contract used a strike of just $0.4708/hr, and this prediction market displays a strong upward trend, moving from a starting price of 8.0% to a current probability of 83.0% , showing the index moved decisively upward within days even off a low base — evidence that the RTX 5090 compute index is capable of fast, large percentage moves. https://octagonai.co/markets/science-and-technology/energy/nvidia-rtx-5090-compute-price-up-or-down-by-apr-10-2026/ - [2026-09-20] The Ornn benchmark data preview shows the broader GPU price index (across H100, H200, B200, A100, RTX 5090) settling at $2.65 per GPU-hour, settled September 20, 2026 , indicating that blended/other-chip compute prices on this same index family are already well above the $2 threshold, which sets a directional tone for the whole index suite. https://data.ornn.com/preview - [2026-09-13] Reporting highlighted "AI Firms Are Reportedly Buying NVIDIA RTX 5090 GPUs in Bulk for Server Use," consistent with a broader pattern in which the Nvidia GeForce RTX 5090 is becoming increasingly difficult to find at retailers, and the reason may go far beyond normal demand from PC gamers — a new report claims that AI companies are buying high-end graphics cards in bulk, sometimes purchasing entire pallets of GPUs at once . https://gaminghq.eu/2026/09/15/rtx-5090-shortage-ai-companies-buy-gpus/ - [2026-09-18] Retail RTX 5090 card prices have more than doubled off MSRP: RTX 5090 is $4,599.99 as of Sep 18, 2026, down 9.8% from $5,099.99 on Aug 19, 2026, tracked between $4,199.99 (Jul 30, 2026) and $7,369.90 (Sep 12, 2026) . Sharply higher hardware acquisition/replacement costs are a structural input that could eventually push cloud providers' rental rates upward as they price in scarcity and capital cost. https://videocardprices.com/card/nvidia-rtx-5090/ - [2026-09-XX] The underlying driver is described as structural, not transient: there are three dominant DRAM manufacturers, and all three have pivoted hard toward high-bandwidth memory and server DRAM for AI accelerators... Analysts at IDC have forecast that AI data centers could consume as much as 70% of the world's memory output in 2026, up from roughly 20% to 30% in 2022 . If this persists into Q4, it argues for continued upward pressure on both card cost and derived compute pricing. https://tech-insider.org/gpu-prices-2026/ - [2026-09-XX] A follow-up analysis notes the cost pass-through is already reaching enterprise GPU servers: the reported 15%+ increase on servers built around Nvidia's AI chips applies to systems shipping starting in early 2027; orders placed now for future delivery are already being repriced to reflect the higher costs , and warns reports out of Taiwan point to a third Nvidia price hike, of up to 30%, that could hit GeForce RTX cards before the year is out — a further hike this would raise the hardware cost base cloud providers must recover. https://tech-insider.org/nvidia-gpu-price-hike-rtx-50-series-2026/ - [2026-09-21] Even so, live cloud on-demand pricing remains far below $2/hr: GPUPerHour shows RTX 5090 on-demand cloud rental currently starts at $0.99 per GPU per hour on RunPod and goes up to $2.40/GPU/hr, based on 8 offers reported in stock by 3 providers , meaning only the single most expensive listed provider (LeaderGPU) currently clears the $2 bar, while the median sits near $0.86–$1.15/hr day-to-day per the site's 14-day history. - [2026-09-20] GetDeploying's aggregate cross-provider tracker likewise shows as of September 20, 2026, the median on-demand price is $0.67 per GPU per hour across 16 providers with a priced on-demand config , well under $2, though this and other trackers show a wide top-end tail (some listings above $2/hr) that a demand shock could pull the average toward. BOTTOM LINE: The strongest YES path is a hardware-cost-driven supply squeeze: the same AI-demand/DRAM shortage that has pushed retail RTX 5090 card prices from $1,999 to $4,600–$7,300+ is structural (IDC/TrendForce data) and could feed through to cloud rental rates if providers reprice to recover capital costs or if AI firms bid up scarce RTX 5090 rental capacity the way they are already bidding up retail cards; current live median rental pricing ($0.67–$1.15/hr) would need to roughly double to triple to clear $2, which is plausible only if the memory-shortage pass-through accelerates sharply into Q4 2026. NOT FOUND: No direct data point showing the Ornn RTX 5090 compute index (the actual Kalshi settlement source) at a level near or above $2/hr at any point in 2026; could not access ornnai.com's historical series directly, and could not find a resolved outcome for the July or August 2026 Kalshi RTX 5090 monthly contracts to gauge recent settlement levels.

- [2026-09-20] Aggregator GetDeploying, tracking 22 GPU-cloud providers (312 price points), put the RTX 5090 cloud rental "median on-demand price" at "As of September 20, 2026, the median on-demand price is $0.67 per GPU per hour across 16 providers with a priced on-demand config." https://getdeploying.com/gpus/nvidia-rtx-5090 - [2026-09-21] The same tracker's live "at a glance" panel showed cheapest in-stock at $0.35/hr and current median at $0.70/hr, roughly one-third of the $2.00 threshold. https://getdeploying.com/gpus/nvidia-rtx-5090 - [2026-09-2x] A second independent aggregator (gpus.io) reported "The RTX 5090 starts at $0.41 per GPU-hour, with a median of $0.80 across 70 available configurations." https://gpus.io/en/gpus/rtx5090 - [~2026-09-17] A third tracker (GPU.ai) stated "RTX 5090 cloud instances on GPU.ai currently start at $0.485 per GPU per hour, billed per second" , describing this as "the live floor price across every provider in our fleet." https://gpu.ai/gpus/rtx-5090 - [2026-09-15ish] AImultiple's cloud GPU index similarly placed the RTX 5090 well under a dollar: "RTX 4090 and RTX 5090 sit below a dollar, at $0.44 and $0.63 respectively." https://aimultiple.com/gpu-index - [2026-09-21] By contrast, Ornn's free public index (the data provider Kalshi uses for settlement) shows much higher rates only for datacenter-class chips: H100 SXM $2.79/hr, H200 $5.07/hr, B200 $7.71/hr, A100 $0.99/hr — the RTX 5090 (a consumer-class card) is not even listed among Ornn's headline free-tier GPUs, consistent with it trading at a materially lower tier than H100. https://data.ornn.com/preview - [2026-09-20] Vast.ai (a major peer-to-peer marketplace) listed RTX 5090 on-demand configs as low as $0.41/GPU/hr and spot as low as $0.20-$0.21/GPu/hr, and RunPod (a leading neocloud) listed $0.69-$0.99/hr — none of the mainstream, high-volume providers approach $2.00. https://getdeploying.com/gpus/nvidia-rtx-5090 - [2026-09-2x] The only outlier configs above $2.00/hr in the GetDeploying dataset came from a single niche provider (Daytona, at $2.52/hr on-demand for certain configs) — an isolated high price, not representative of the market median, which a volume-weighted or median-based index would filter out. https://getdeploying.com/gpus/nvidia-rtx-5090 - [2026-07-15] Reporting on Kalshi's launch of these GPU-compute markets noted the settlement mechanism uses Ornn's index, which is a "commodity class comparable to energy and metals... giving traders a transparent benchmark rather than relying on anecdotal marketplace pricing" — implying the settlement will reflect broad market medians, not cherry-picked high quotes. https://cryptobriefing.com/kalshi-gpu-compute-price-prediction-markets/ - [2026-07-15] The same coverage flagged a real precedent for cloud-GPU prices falling rather than rising: "NVIDIA's B200 saw its per-hour compute rate peak at $6.11 on May 30, 2026. By June 21, it had dropped to $4.22. That's a roughly 31% decline in less than a month." — showing GPU compute markets can swing down quickly even amid AI demand narratives, undercutting the case for a sustained tripling of RTX 5090 rates by December. https://cryptobriefing.com/kalshi-gpu-compute-price-prediction-markets/ - [2026-09-xx] Note the key resolution-criteria distinction: retail RTX 5090 hardware prices have surged to $4,000-$5,000+ due to memory/AI-driven shortages, but this is a separate market (physical card purchase price) from "compute per hour," which is the cloud-rental/inference rate — the two should not be conflated, and the hardware-price surge narrative does not by itself imply hourly rental rates are near $2.00. BOTTOM LINE: Every independent cloud-GPU price tracker checked in September 2026 puts the RTX 5090's on-demand median rental rate between roughly $0.63 and $0.80 per GPU-hour — less than half of the $2.00 threshold — with only isolated niche-provider quotes exceeding $2, which a volume-weighted settlement index would not let dominate the average; absent a sudden multi-fold supply shock or provider consolidation, the current trend and methodology strongly favor a NO resolution. NOT FOUND: The actual Kalshi market page for the December 2026 RTX 5090 monthly contract (kxrtx5090ms-26dec) could not be fetched (domain access blocked), so current market-implied odds/trader positioning were not directly observed; similarly, Ornn's specific RTX 5090 index page/historical series (as opposed to its H100/H200/B200/A100 headline data) could not be accessed directly to confirm the exact settlement-grade RTX 5090 figure.