According to a16z, citing Ornn Data, Nvidia GPU rental prices have surged this year: H200 rates are up 160% and B200 up 120% since January, while RTX PRO 6000 rose about 65% and H100 35%. Even the six-year-old A100 costs more to rent than at the start of the year. a16z said rental prices across every GPU model it tracked were 9% to 143% higher than a straight-line six-year depreciation model would imply, suggesting chips are holding value better than expected. Token consumption on OpenRouter has climbed sharply since spring, though a16z cautioned this is only a partial view of AI demand. Meanwhile, AI-linked stocks sold off over the past two weeks, and NVDA is down 7.6% this month, closing at $210.96 on Friday.
Summary written by Upside AI from the original article. Read the source for full details.
Read the original on Yahoo Finance ↗
Keywords
NVDAsemiconductorsartificial intelligencecloud computing
Open a keyword to see other news on the same topic.
Related tickers and likely impact
NVDA
▲ Likely positiveNvidia
Relevance: medium
| Market | US stocks |
|---|---|
| Sector | Information Technology |
| Industry | Semiconductors |
| Market cap | $5.1T |
| Close on 10/23/2026 | 211.00 |
| Change that session | -0.57% |
Impact is scored by AI from the headline and may be wrong. It is not investment advice.
Related news
- Broadcom's Time is Nearing as Neocloud Risk Jolts Nvidia
The author says they bought more Broadcom shares after the stock fell 9.07% in the week ending October 16 to $328.75, citing its role co-designing custom AI chips for large customers. AI semiconductor revenue reached $16.70 billion in fiscal Q3, up 221%, and management guided Q4 to $21.70 billion. The company says it has already secured supply for fiscal 2027 and 2028, generated $13.7 billion in free cash flow, and pays a $0.65 quarterly dividend. The article flags risks such as customer concentration, while noting that NVIDIA has approached insurers about covering losses on loans to neocloud customers. Broadcom reports again on December 9, 2026, against guidance for $34.80 billion in revenue.
- Nvidia wins CPU business from Modal, previously exclusive to AMD and Intel
Modal, a San Francisco-based AI cloud startup that previously ran its AI agent workloads only on Intel and AMD processors, will begin using Nvidia's new Vera CPU. Early tests showed the Nvidia chip doubled work capacity for AI agents, and Modal plans to run it alongside Intel and AMD chips, assigning each task to the most efficient hardware, though integration requires substantial software changes. Modal recently raised $750 million in a round led by Accel at a $15.75 billion valuation and expects $1 billion in sales over the next year, with usage tripled and revenue up more than 15 times year-over-year. Nvidia projects at least $20 billion in Vera chip revenue this fiscal year in a market it estimates at about $200 billion, and says OpenAI, Anthropic, Oracle and Perplexity also plan to use the chips.
- As Big Tech Designs Its Own AI Chips, the Companies Behind the Scenes Win
The article argues custom AI processors are becoming more important alongside GPUs, as they can run specific tasks faster and more efficiently, naming Broadcom, Marvell and TSMC as key beneficiaries. Google Cloud sales rose 82% to $24.8 billion in the second quarter, and Alphabet signed a multiyear custom processor deal with Broadcom running through 2031. Broadcom management expects AI semiconductor revenue of $56 billion this year and more than $100 billion next year. TrendForce estimates custom processors will grow about 45% this year versus 16% for GPUs, while AMD's deal with Meta could be worth up to $60 billion. The article says TSMC benefits regardless of demand for GPUs or custom chips, as it makes about 73% of the world's processors and an estimated 90% of the most advanced ones.
- Chinese Tech Firms Bought 1 Million Nvidia Gaming Chips This Year. Here's Where It Shows Up
According to the Financial Times, Chinese tech groups have bought more than 1 million Nvidia gaming and workstation chips this year to cover a shortfall in AI computing power. These products are not blocked by U.S. export rules; they are slower than Huawei's Ascend 910C at most tasks but adequate for simpler AI work. Domestic Chinese AI chip supply could reach about 3.5 million units this year, while demand is expected to far exceed 5 million. Nvidia's revenue from customers headquartered in China, including Hong Kong, reached $7.88 billion in the most recent quarter, roughly double a year earlier, yet its guidance still assumes no China data center compute revenue. Oracle is named as the clearest U.S. beneficiary, as Tencent reportedly agreed to lease access to about 100,000 advanced AI chips through Oracle's Southeast Asian data centers.
- Microsoft reveals Maia 300 AI chip, as it works to reduce dependence on Nvidia, AMD
Microsoft has unveiled the Azure Maia 300 AI chip to reduce its reliance on Nvidia and AMD, to be deployed at multigigawatt scale from next year alongside Maia 200 and Maia 100. Microsoft says Maia 300 offers 40% more tokens per megawatt than leading GPU-based systems, and a 96-chip rack delivers seven times the rack-scale token throughput of Maia 200. Maia 200 already powers workloads including Microsoft 365 Copilot and OpenAI models, with 1.7 times more tokens per watt than other GPU systems. Peers such as Amazon, Google, Meta, OpenAI and Anthropic are also building custom chips to avoid overdependence on Nvidia and AMD. The article notes that Nvidia regularly generates over 50% of its quarterly revenue from hyperscalers, and that for now hyperscalers still need to buy Nvidia's most in-demand chips.
- How AI Chips Will Power Google Cloud Revenue In 2027 And Beyond
Related markets
Latest news
- Cameco Stock Is Down 42% From Its Record. Here's How I'd Buy It Around Its Oct. 30 Report.
Cameco shares closed at $78.60 on Oct. 23, their lowest close since September 2025 and about 42% below the record close of $134.09 on Jan. 28. Long-term uranium prices reached $96.50 a pound at the end of September, above the 2007 peak of $95, but Cameco's realized price was only $67.79 in the second quarter because many contracts were signed earlier. Second-quarter adjusted EPS fell 75% mainly because a Czech reactor project's roughly US$170 million revenue contribution a year earlier did not recur, while deliveries dropped 18%. Cameco still expects 2026 deliveries of 29 million to 32 million pounds, and trades at about 40 times next year's expected earnings versus about 19 times for the S&P 500. The author sees the sell-off as more about interest rates than Cameco itself, but considers the valuation still demanding and suggests accumulating shares in stages around the report.
- Gulf Nation Sends Positive News Across All of Asia
Saudi Arabia plans to store crude oil in the Jurong Rock Caverns, an underground rock cavern system 150 m deep in Singapore, from the second half of 2027, with Saudi Aramco set to sign a memorandum of understanding with EMA and JTC. The move aims to bring supply closer to Asian buyers after the US-Israel conflict with Iran disrupted energy shipments through the Strait of Hormuz, when about two-thirds of Asia's oil imports previously came from the Middle East. The caverns can hold up to 1.47 million m3 of liquid hydrocarbons, roughly 9 million barrels of crude and condensate. Aramco already stores oil in Japan and South Korea, while Japan is negotiating to expand regional oil reserves through the POWERR Asia initiative, backed by a US$10 billion financial package.
- Honeywell Technologies Launches Forge Plant Cognition with Dangote, Built with NVIDIA AI and Dell Technologies Capabilities
Honeywell Technologies has launched Forge Plant Cognition, an industrial AI platform built with NVIDIA AI and accelerated computing and Dell Technologies' AI infrastructure, designed to help refining, petrochemical and LNG operators get more value from existing assets. The platform is launching first at Dangote's Lekki refinery in Nigeria, where preliminary estimates show AI-enabled optimization could deliver more than 29 cents a barrel in operational benefit, or more than $60 million a year in gross benefit at full capacity. The solution aims to identify bottlenecks, detect equipment issues early to reduce unplanned downtime, and embed process expertise into AI workflows to support faster, more consistent decisions.
- An EV Business Revival and Potential SpaceX Merger: Is Tesla Stock a Buy?
Tesla's EV business appears to be rebounding, and a possible merger with SpaceX (SPCX) is drawing more attention, though Musk says he is not part of any merger talks. In the third quarter, deliveries fell 5% while production rose 4%; auto revenue dropped 3% to $20.56 billion, while active FSD subscriptions rose 67% to 1.74 million. Service revenue jumped 33% to $4.6 billion, and adjusted EPS slipped 4% to $0.48, still above the $0.43 analyst consensus. Tesla is also developing the Cybercab robotaxi, Optimus robots and the Semi truck, and the author judges this a favorable time to add the stock.
- Bitcoin Drops Below $80,000
Bitcoin slipped below $80,000 around midday Vietnam time, down nearly 2% over 24 hours, and at one point dipped to $79,300 before buyers offered some support. Investors cut risk ahead of this week's Fed meeting and an Asian AI-linked tech stock selloff, pushing most major cryptocurrencies lower: Ether fell nearly 3% to below $2,300, XRP dropped to $1.28 and Solana hovered near $100. BTSE's Jeff Mei said the key is how strongly the Fed chair signals a possible rate hike in December, as record-high oil prices, escalating US-Iran tensions and inflation could push the Fed toward a firmer message. CoinDesk analysts said Bitcoin needs to recover $82,500, a level it lost last weekend, for buyers to treat the drop as a temporary correction.
- SWI Compute Emerges from Stealth to Turn Gigawatts of Power Into NVIDIA AI Factories, Starting in Europe
SWI Compute, which has emerged from stealth, will build and operate NVIDIA AI factories, launching first in Europe. The company is backed by 3.6 gigawatts of power capacity that parent SWI Group, listed on Euronext Amsterdam, spent six years securing across Europe and the US. It became a Preferred NVIDIA Cloud Partner in August, ordered approximately $1 billion of NVIDIA systems, and has finalized contracts with its first customers. Its first large UK AI factory starts with 30 MW, close to 7,000 Rubin GPUs and 3,500 Vera CPUs in Vera Rubin NVL72 racks expected online by the end of 2027, as the first step toward a campus planned to reach 530 MW and 2.3 GW of capacity by 2029. Delta Computers is the systems integrator, and the partners are set to deliver more than 8,000 GPUs.
Get the briefing on Telegram or Zalo
Join the Upside Telegram channel or Zalo group for the Vietnam stocks and gold briefing at 8:00 and 19:00 (Vietnam time) on trading days, plus a weekly wrap on Saturday morning.