B200

Bitget

2026-09-08 06:11:39

Bitget Launches First Pre-IPO Perpetual Contracts for H100 and B200 GPU Rental Prices

Bitget has launched the first pre-IPO perpetual contracts for retail investors, H100 and B200, pegged to the rental price indices of NVIDIA's H100 and B200 GPUs. Quoted in USD per GPU hour, the products use USDT settlement, support up to 10x leverage, and offer 24/7 trading, enabling retail users to participate in AI computing price discovery.

1040

data center c

2026-09-03 06:43:09

Social Graph VC: Data Center CapEx to Surpass $1 Trillion by 2026, CME to Launch GPU Futures

Venture capital firm Social Graph VC released a primer on the computing market, forecasting that data center capital expenditure will exceed $1 trillion in 2026, roughly double the combined spending of the four major hyperscale cloud providers in 2025. Jensen Huang predicts annual spending could reach $3-4 trillion by the end of the decade and $10 trillion by 2031. AI investment currently accounts for about 0.9% of global GDP, rising to 1.4% by 2028. The training costs for Fable 5 and GPT-5.6 are each around $120 billion, with power consumption below 2 GW. CME plans to list cash-settled H100 and B200 monthly lease futures on NYMEX on October 5, 2026, pending regulatory review, referencing the Silicon Data Index. ICE also announced GPU futures based on the Ornn Index.

950

South Korea

2026-08-31 14:19:32

South Korea plans free unlimited-token AI service for all citizens

The South Korean government said it will provide every citizen with an AI service that is free to use and comes with no token cap, according to Techub, citing Decrypt. The service is set to be operated by three consortiums, which will share up to 512 Nvidia B200 chips supplied by the state. Based on the information released so far, the project is expected to enter beta testing in September. No additional rollout details, pricing structure beyond the free access model, or participant names for the three consortiums were disclosed in the source material. The announcement centers on public access, operating structure, computing resources, and the planned beta timeline.

690

CME

2026-08-11 15:02:01

CME and Silicon Data to launch compute power futures on Oct. 5

CME and Silicon Data have set Oct. 5 for the launch of compute power futures, according to a BlockBeats report published on Aug. 11. The initial lineup will include two contracts: the Silicon Data H100 Rental Index Futures and the Silicon Data B200 Rental Index Futures. Both products are designed to track indexes that measure the hourly rental cost of GPUs. The update identifies the launch date, the partners involved, and the names of the first two contracts, while indicating that the underlying benchmarks focus on GPU leasing prices on an hourly basis.

1810

CME

2026-08-11 15:23:03

CME to List H100 and B200 Lease Index Futures on New York Commodity Exchange

According to Odaily Planet Daily, CME Group has announced that it will list futures contracts for H100 and B200 lease indexes on the New York Commodity Exchange. These two contracts will track an index measuring the hourly rental cost of GPUs.

1660

AMD

2026-08-04 11:15:08

Wafer says Kimi K3 fits on one 8x AMD MI355X server, while B200 needs 16 GPUs across two nodes

Wafer AI said it deployed Kimi K3 on AMD’s MI355X, claiming the model can run on a single server with eight GPUs, while the same workload requires 16 NVIDIA B200 GPUs spread across two servers because of memory limits. In Wafer’s test with 1,024 input tokens and 400 output tokens, the MI355X setup delivered 952 tokens per second of aggregate throughput and 118 tokens per second for single-user generation. The article says that, on a single-node basis, this works out to roughly 3.8 times the throughput of the 16-card B200 deployment’s average per-node figure, while B300 still leads on absolute speed at 1,568 tokens per second and 172 tokens per second for single-stream generation. Wafer also published a cost comparison using hourly prices of $2.5 for MI355X, $4.25 for B200, and $6 for B300, under which MI355X offered about 48 tokens per second per dollar, ahead of B200’s roughly 7 and B300’s roughly 33. The report also highlights ROCm software progress, saying Kimi K3 was largely able to run directly on MI355X, with follow-up work focused on a small number of compatibility issues and performance tuning, including speculative decoding fixes and a prefill optimization that reduced cold-start latency on a 172,000-token task.

3010

South Korea

2026-07-22 09:16:58

South Korea launches national AI tender with 512 NVIDIA B200 GPUs and a domestic-model quota

South Korea’s Ministry of Science and ICT has opened bidding for its “AI for All” project, a national initiative that aims to deliver free, unlimited-use AI chatbot access and a public-service AI agent to the general population. The government plans to pick two to three private-sector operators, with bids due by Aug. 11, selection expected by late August, a beta launch set for September, and a nationwide rollout scheduled for December. Instead of handing out cash upfront, the government will subsidize the project with 512 NVIDIA B200 GPUs. If two companies are chosen, each can receive 256 GPUs; if three are selected, the top-ranked bidder gets 256 while the other two receive 128 each, subject to final adjustment based on demand and evaluation results. Government support may continue through Dec. 31, 2030, with operating costs covered directly from the state budget starting in 2027. The tender also sets a strict usage rule for local AI developers. More than 50% of output must come from qualified domestic foundation models, and more than 30% must come from other Korean companies’ models, measured by output tokens. Foreign models are allowed only for functions domestic systems cannot handle, such as certain high-difficulty reasoning or multimodal tasks, and those usages are excluded from subsidies.

1790

Kalshi

2026-07-15 05:00:20

Kalshi launches GPU price prediction markets for Nvidia B200, H200 and A100

Prediction market platform Kalshi said on July 15 that it has launched a GPU compute forward curve covering Nvidia’s B200, H200 and A100 chips. The new market is designed to show expected future compute pricing based on market forecasts. The launch brings GPU pricing into a prediction-market format and centers on three Nvidia products named in the announcement. No additional details on contract structure or timing beyond the announcement were disclosed in the source material.

1270