GPU rental prices (H100) end of July?
I think a July 31 print below $2.00 is unlikely, though not impossible. The most recent Ornn readings and the broader H100 pricing backdrop both still sit noticeably above the threshold, so I lean No with only a small Yes tail.
Analysis
The most recent Ornn public readings I could find show H100 at $2.64/hr on the compute page crawled 4 days ago, and $2.99/hr on the homepage crawled 2 weeks ago. Both are materially above the $2.00 cutoff, so a July 31 Yes would require a meaningful late-month decline rather than a marginal tick down.
The broader pricing backdrop is mixed but still not comfortably below the threshold. A weekly market recap said H100 rental prices landed in the $2.30-$2.60 range by June 30, while recent provider comparison pages still show the cheapest fixed on-demand H100 around $2.19 and mainstream public cloud rates much higher, with AWS p5 H100 at about $4.33 per accelerator and Google Cloud’s H100 on A3 Mega at about $11.76 per GPU-hour.
That combination argues for a low Yes probability: the index has already softened from the high-$2s but not enough to imply a clean break under $2, and the market would need another roughly 25% drop from the latest Ornn public reading to settle in Yes territory. The thin-but-real lower-priced marketplace offers keep the tail alive, but the current center of gravity still looks closer to the mid-$2s than to $1.90 or below.
Arguments
For
- Arguments for Yes: Continued substitution toward newer accelerators and competitive marketplace pricing could push transacted H100 rates lower before month-end.
- Arguments for Yes: Some public comparisons already show H100 offers near $1.73-$2.19, so the floor is not far from the target in the cheapest segments.
Against
- Arguments against Yes: The latest Ornn readings are still in the mid-to-high $2s, so the benchmark has not yet demonstrated a sustained move below $2.00.
- Arguments against Yes: Official hyperscaler pricing remains far above $2.00, which suggests the blended index would need a strong further repricing to cross the line.
Key drivers
- Recent Ornn public prints around $2.64-$2.99 anchor expectations well above the cutoff.
- Cheaper marketplace inventory and provider competition can still compress the blended benchmark if demand weakens.
- The final settlement depends on the July 31 daily print, so late-month volatility matters more than the intramonth average.
Risk factors
- A late-month supply squeeze or usage surge could keep the finalized print above $2.00.
- The index could fail to move enough in time because it is still anchored above $2.50 in the latest public Ornn read.
- A sharper-than-expected repricing from cheaper provider inventory could pull the daily settled value under the threshold.
Scenarios
Best case
Late-July supply loosening and heavy migration to cheaper capacity pull the finalized July 31 print into the high-$1.80s or low-$1.90s, producing a Yes.
Most likely
The benchmark eases modestly but not enough, ending somewhere around $2.10-$2.40 and resolving No.
Worst case
The index stabilizes in the mid-$2s or rebounds, leaving the finalized print comfortably above $2.00 and resolving No.
More from this day
- techPolymarketEnded
Best AI model on July 18?
AI14%MKT93%Edge-79HypedI think claude-opus-4-6-thinking is materially less likely than the market price suggests to be the overall #1 model on July 18, 2026. The current leaderboard evidence points to Claude Fable 5 as the incumbent leader, so I put Yes at 14%.
- HealthKalshi2y
What will the average number of measles cases be during Trump's term?
AI80%MKT28%Edge+52Hidden GemAssuming the market's binary question is whether the 4‑year (2025–2028) average annual measles cases will be >= 2,000, I assess an 80% probability that the average will meet or exceed that threshold.
- EconomicsKalshi2y
Will there be a Trump economic boom?
AI8%MKT51%Edge-43HypedIndependent assessment: very unlikely — I assign an 8% chance that any U.S. quarterly real GDP (annualized) will exceed 5% in Q1 2025–Q4 2028. The data and policy environment point strongly against a sustained or one‑off 5%+ quarter, though low‑probability shocks could still produce a spike.