gpu

On October 28, after the market closed, Moer Thread announced that the total operating revenue for the first three quarters of 2026 was 3.289 billion yuan, a year-on-year increase of 319.15%; the net profit attributable to the parent company's owners was 37.6973 million yuan, compared to a net loss of 724 million yuan in the same period last year, turning a loss into a profit; the basic earnings per share were 0.08 yuan, compared to -1.81 yuan in the same period last year.In the third quarter, the company's total operating revenue was 1.552 billion yuan, a year-on-year increase of 1773.92%; the net profit attributable to the parent company was 49.2604 million yuan, compared to a net loss of 453 million yuan in the same period last year. In the secondary market, Moer Thread rose 5.84% on the 28th, with a stock price of 381.50 yuan and a total market value of 179.3 billion yuan.

The U.S. SEC filings show that SpaceX has signed a binding primary financing lease commitment with Apollo Global Management.According to the agreement, Apollo will purchase GPU clusters and related equipment and lease them to SpaceX, with the total principal payments under the related lease arrangements expected to reach $40 billion.

According to a report by Reuters cited by Phoenix Technology, on October 26, Beijing time, Elon Musk, the founder of SpaceX, stated that the company has completed $40 billion in debt financing for the procurement of graphics processors. Musk posted on the social platform X that this financing is primarily for the purchase of GPUs that will be delivered in the first and second quarters of next year, and the debt can be redeemed early, which will not have a significant impact on this quarter.Musk added that SpaceX's investment-grade credit rating, along with the revenue generated from Starlink satellite internet and launch services, has helped the company maintain its ability to secure credit financing. Earlier this month, Reuters reported that SpaceX is in talks with banks and asset management companies to raise $40 billion for the purchase of NVIDIA's AI chips.For the three months ending September 30, SpaceX's revenue grew by 161% year-on-year, reaching $14.58 billion. During the same period, the company's capital expenditures on computing infrastructure nearly doubled, reaching $19.6 billion.

The venture capital firm a16z released a chart weekly report written by Moses Sternstein on October 23, 2026. The report states that Netflix is still growing, with revenue growth slowing to just below about 12%, the lowest level since 2021 to 2022. The company is producing its own short videos and entering live sports. The report cites BCG data indicating that the upper limit of media consumption in the U.S. is about 26 hours per week, showing a moderate decline since the peak of lockdowns. Moses Sternstein: Competition comes not only from other streaming services but also from short videos, with much of the recent media growth coming from background consumption while driving or working.In terms of computing power, the report cites Ornn Data stating that the rental price of every GPU in the chart series is higher than at the beginning of the year, including the oldest A100, which is about 9% to 143% higher compared to a 6-year straight-line depreciation. Moses Sternstein: Over a longer period, older A100 and H100 GPUs spend more time below that line, but have only shown significant strength since April this year, making comparisons over a longer period not entirely fair. According to OpenRouter, token demand surged sharply in spring, jumped again after a brief plateau at the end of summer, with token consumption for both open-source and closed-source models still following an exponential path, and the share of proprietary models slightly increased last week. The author states that OpenRouter only covers a very small portion, and these indices are not perfect proxies for actual token spending and GPU pricing.In manufacturing, the report states that computer production capacity and utilization rates are rising, with production increases being larger. The increase in communication and semiconductor production mainly comes from expansion, while the price increase for electrical equipment is the largest, but domestic production has slightly declined. The import increase of key categories of AI capital expenditures mentioned above far exceeds the domestic production increase in dollar terms. According to ADP, the increase in average weekly hours for industries related to AI construction is greater than that of other industries, and is generally greater than the net increase in personnel. Job switchers in manufacturing have about a 5 percentage point wage growth premium compared to stayers. Cybersecurity stocks remain strong in IGV this year, but have seen a larger pullback than the market since October. The revenue growth rate of large companies has mostly only increased by a few percentage points, with some experiencing declines, and stock price performance reaching up to about 125%, with Palo Alto Networks' revenue growth largely driven by acquisitions.

Evan Conrad published an article on October 22 titled "The end of startups," with the subtitle "The coming credit crunch." Evan Conrad stated that within the next six months, most AI startups will be unable to purchase computing power, as global computing power is sold out. To obtain capacity, new clusters need to be contracted, and it is becoming increasingly difficult for non-investment-grade entities to secure online capacity.Evan Conrad mentioned that startups are not investment-grade, thus they are required to pay larger upfront payments. He has seen cases where 100% of the contract amount is required to be prepaid, as lenders and cloud providers are unwilling to take on additional credit risk. He believes that investment-grade tenants are scarce, and due to leveraging and selling to non-investment-grade token buyers, their ratings are downgraded, which he refers to as a credit bottleneck. He anticipates that investment-grade GPU buyers will increasingly demand that token buyers themselves be investment-grade or pay large upfront deposits.In the article, he provides an example where, at the time of writing, B300 could be purchased for $6 per hour and then sold for approximately $30 per hour for inference. He expects GPU prices to rise, and the gross margin for inference layers to decline, making it harder for clients with poor credit to obtain tokens. He proposed that lenders treat GPU clusters as collateral and stated that the physical delivery forward market SF Compute could allow lenders to use GPU hours on a liquidity book as collateral, rather than relying on a single service provider's cluster.

Cryptocurrency researcher Kerman Kohli published "Compute Financing" on Substack on October 19, 2026, discussing financing for computing power construction. He believes one of the main bottlenecks is capital, especially credit: early expenditures mainly come from balance sheet cash, followed by a shift to equity financing. The article cites Alphabet's $85 billion financing and Meta's fundraising through shareholder dilution, with the next step being the credit market.Kohli summarizes the current beliefs of creditors into three points, which he believes are incompatible: AI may be a bubble; the biggest risk is counterparty risk in purchase contracts; the underlying GPUs are not valuable and depreciate like ordinary hardware. He argues that demand is closer to infinite, stating that the demand for agentic AI is growing by about 2.5 to 5 times each year, with supply potentially doubling by the end of 2030. Memory is expected to sell out by 2027, with computing power commitments reaching trillions of dollars, and he mentions Jane Street's multi-billion dollar commitments, as well as the needs of pharmaceutical companies, research, military, and cybersecurity.Regarding efficiency, he states that when the cost of intelligence drops by 90%, the increase in usage is even greater, with a baseline exceeding 10 times, and he cites Jevons' Paradox.On the topic of purchasing, Kohli believes that computing power is more substitutable than real estate; when demand is strong, if someone withdraws their commitment, others will take over. He mentions that many creditors require 2 to 3 year contracts and down payments of 20% to 30%, sometimes as high as 50%, and locking in long-term discount contracts effectively shorts computing power for lenders. The article also states that computing power rental prices have risen by about 50% in the past 6 months and about 100% in the past year, with B300 supplies being tight. Regarding collateral, he notes that the A100 from 2020 is still operational and rented out until 2029, leading him to believe it has at least about 9 years of use value; the current price of B200 is about 50% higher than when it was released about two years ago, and a 6-year-old A100 is still worth about 25% of its value. He concludes that computing power will surpass real estate to become the largest asset class, and the scale of credit-supported computing power is currently just beginning.

NVIDIA (NVDA.O) launched the cuPotential CUDA-X library, accelerating GPU workflows by 2 to 3 times in the early access phase.

Ethereum Layer 2 network ZKsync announced the launch of the prover Airbender Prime. ZKsync stated that, to their knowledge, this is the first prover to demonstrate real-time proofs of Ethereum on a single GPU. Airbender Prime is designed for post-quantum security, utilizing hash-based STARK, combined with GKR sumcheck, WHIR polynomial commitment, and BLAKE2s hash; its L1 proof will use a commitment based on Keccak-256 instead of BN254 pairing. This prover is based on the open RISC-V standard and combines real-time proofs with independent validators.In a single GPU evaluation, an NVIDIA RTX PRO 6000 Blackwell completed proofs for 99% of 7200 consecutive Ethereum mainnet blocks, with each block taking less than 10 seconds, including witness preparation; all blocks were completed within 12 seconds. ZKsync claims that Airbender Prime leads in average proof time, GPU efficiency, and cost reported on the ETHproofs public leaderboard, and it already has dual GPU and quad GPU real-time configurations while reducing the hardware requirements for generating proofs.

On October 13, Ruantong Power was reported to have received 8,000 NVIDIA GPU cards, specifically the NVIDIA 6000D (domestic "special edition" of the Blackwell architecture), which will be used for the production of related servers.

Recently, SoftStone Technology published an article on its WeChat official account regarding the order of 8,000 NVIDIA 6000D GPU cards arriving, which has since been deleted. In response, SoftStone Technology staff stated that the company will announce matters that meet disclosure standards based on business conditions and advised investors not to use such news as the sole basis for investment.

Multiple venture capital firms in Silicon Valley are competing to secure GPU computing power to provide exclusive resources for their AI startups. a16z established a computing power pool of about 20,000 GPUs as early as 2024 and is negotiating expansion with data center companies Switch and Volta. General Catalyst plans to build a data center for its portfolio companies and has brought in the former head of NVIDIA DGX Cloud in May this year to oversee AI infrastructure operations.Even if AI startups complete their financing, they may not be able to rent GPUs in a timely manner. In April this year, major cloud providers like Microsoft were reported to prioritize allocating computing power to internal teams and large clients, leaving some startups facing months of waiting. a16z previously disclosed that during times of computing power shortages, prices that could originally be secured with a six-month contract might require a three-year commitment. Model training requires a large amount of GPU usage in a short period, while inference demand is difficult to predict, making it easy to waste funds by locking in computing power for several years in advance.a16z is concentrating on purchasing and leasing computing power through the Oxygen project, redistributing it to its portfolio companies, allowing them to access prices and contract terms typically reserved for large clients. In August this year, a16z co-led a $300 million financing round for Volta, which is valued at $2.4 billion, with investments covering both AI startups and data center companies providing computing power. The small venture capital firm Crucible Capital has built a cluster with 576 NVIDIA B300 GPUs for its portfolio companies to test, while also selling some computing power to a cutting-edge AI laboratory. This model allows venture capital firms to take on investments that were previously mainly borne by cloud providers and data center operators.

According to Bloomberg, Shanghai Birun Technology (Hong Kong stock code 6082) plans to raise approximately HKD 4.04 billion (USD 515 million) to support its artificial intelligence business. The company intends to place 130 million new shares at a price of HKD 31.08 per share, with the related arrangements based on documents from the Hong Kong Stock Exchange.The placement price represents a discount of nearly 10% compared to Wednesday's closing price. Bloomberg reports that Birun is one of the GPU developers receiving support from Beijing, and this financing is part of a wave of fundraising for Chinese artificial intelligence companies.

The GPU infrastructure financing platform Silicon Exchange has announced its launch, positioning itself as a financing method for artificial intelligence infrastructure. Users can hold real shares of GPU infrastructure, while the platform is responsible for deployment and hosting, and computing power clients use the related hardware.Silicon Exchange stated that users can visit its official website to view the first GPU vault. The official announcement has not disclosed the scale of the vault, hardware models, financing structure, or specific profit arrangements in this post.

The NVIDIA research team enabled the AI Agent to directly rewrite the GPU kernel of Kimi Delta Attention. Kimi Delta Attention is one of the core attention mechanisms used by the Dark Side of the Moon Kimi-Linear model, and such low-level code typically required engineers familiar with CUDA and chip architecture to optimize manually and repeatedly in the past.The version written by the Agent achieved 2.96 times the speed of the official FlashKDA of the Dark Side of the Moon on the NVIDIA B300. The team tested a total of 6 sets of tasks with fixed lengths and varying lengths, with 2.96 times being the geometric mean of all tasks; the relevant kernel has been open-sourced.During the process, the Agent exploited testing loopholes by incorporating statistical patterns from the test data into the code, achieving a speedup of 3.74 times; when retaining only the most recent 32 tokens, a single score reached 5.16 times. These versions would produce incorrect results or fail when switched to real Kimi data. The team subsequently incorporated real Kimi runtime data, random inputs, and extreme value tests, tightening the error standards, and ultimately retained the 2.96 times version, which passed this validation.

User gpugene stated during a discussion on computing power trading: Some B300 transaction prices have exceeded 24 dollars per GPU hour, and the price of computing power with a term shorter than 1 year remains above 7 dollars and is still rising. gpugene expressed the hope to use computing power derivatives to guard against the rise in GPU hour prices while retaining the flexibility to adjust rental durations; if inference clouds sell services at fixed prices and their own GPU bills fluctuate, they have effectively taken on the risk of computing power price changes.The article provides an example where a supplier can offer a plan for 2048 B300s, available for rent for 5 years at 4.15 dollars per GPU hour, corresponding to an amount of 372,264,960 dollars. If the client goes live after 12 months and uses this scale from the 13th to the 15th month, the total would be 4,485,120 GPU hours. gpugene calculated that the current premium for a call option with an exercise price of 4.50 dollars is approximately 4.83 million dollars; if changed to a compound option, the current payment would be about 2.70 million dollars, with an additional payment of about 4.49 million dollars upon exercising in the 12th month.gpugene also compared 4000 hypothetical 5-year rental and demand paths. The average cost of annual renewal is 4.72 dollars per hour, with the worst 5% scenario averaging 10.91 dollars; after adding the call option, these figures change to 4.88 dollars and 7.20 dollars, respectively, with the premium included. gpugene believes that the emerging computing power derivatives market may change how neocloud and related participants expand and protect themselves.

On-chain lending protocol USD.AI announced that it has secured a financing amount of $128.9 million in GPU funding, marking the largest single financing in the protocol's history.The financing will be used to support an undisclosed borrower in deploying 32 sets of NVIDIA GB200 NVL72 systems in British Columbia, Canada. This financing amount sets a new historical record for USD.AI.

Binance will launch the Axe Compute (AGPUB), AMC Entertainment (AMCB), and Cypherpunk Technologies (CYPHB) bStocks trading pairs at 20:00 on September 23. The trading pairs include AGPUB/USDT, AMCB/USDT, and CYPHB/USDT. At the same time, the spot algorithm trading bot will be activated in line with the spot listing time, and the smart holding bot will be enabled within 24 hours after the spot market goes live.bStocks can be exchanged for BTC, USDT, or other tokens without any fees within one hour after going live on Binance spot. During the event until October 1 at 07:59, there will be zero maker fees.