BTC $64,545.83 -0.09%
ETH $1,810.16 -0.28%
BNB $593.50 +0.46%
XRP $0.9744 +2.14%
SOL $74.52 +4.06%
TRX $0.3078 +0.84%
DOGE $0.0594 +1.21%
ADA $0.1530 +0.79%
BCH $209.85 +1.09%
LINK $9.22 +1.37%
HYPE $68.97 +0.04%
AAVE $111.80 +2.73%
SUI $0.7291 +2.14%
XLM $0.1402 +2.51%
ZEC $959.31 +6.36%
AAPL $350.19 -0.97%
AMZN $259.52 +0.22%
GOOGL $283.32 -0.48%
MSFT $508.50 -2.24%
META $620.24 -0.90%
NVDA $160.63 +2.54%
TSLA $383.99 +0.90%
SNDK $825.28 +1.56%
INTC $56.12 +3.71%
SPCX $142.79 +3.64%
MU $734.95 +1.43%
AMD $515.04 +0.85%
BTC $64,545.83 -0.09%
ETH $1,810.16 -0.28%
BNB $593.50 +0.46%
XRP $0.9744 +2.14%
SOL $74.52 +4.06%
TRX $0.3078 +0.84%
DOGE $0.0594 +1.21%
ADA $0.1530 +0.79%
BCH $209.85 +1.09%
LINK $9.22 +1.37%
HYPE $68.97 +0.04%
AAVE $111.80 +2.73%
SUI $0.7291 +2.14%
XLM $0.1402 +2.51%
ZEC $959.31 +6.36%
AAPL $350.19 -0.97%
AMZN $259.52 +0.22%
GOOGL $283.32 -0.48%
MSFT $508.50 -2.24%
META $620.24 -0.90%
NVDA $160.63 +2.54%
TSLA $383.99 +0.90%
SNDK $825.28 +1.56%
INTC $56.12 +3.71%
SPCX $142.79 +3.64%
MU $734.95 +1.43%
AMD $515.04 +0.85%

delivery

All
Article
Flash

first_img AI data centers are turning to second-hand gas turbines, with new machine delivery cycles lasting up to 6 years

According to a report by Decrypt, due to the long delivery cycle of new gas turbines, AI data center developers are turning to second-hand gas turbines and retired jet engines. Data from the energy consulting firm Wood Mackenzie shows that the average delivery time for new gas turbines in the United States has reached 6 years, with related delivery discussions focusing on 2030 and 2031. Reuters reports that global gas turbine orders have exceeded 100 gigawatts, while manufacturers' annual production capacity is estimated to be only 60 to 70 gigawatts. Siemens Energy, GE Vernova, and Mitsubishi Heavy Industries have all stated that order growth exceeds production capacity.One response is to retrofit gas turbines based on aircraft jet engine technology. AI infrastructure company Crusoe ordered 13 units from ProEnergy in Missouri in April, each with a rated power of 50 megawatts, totaling 650 megawatts, for future data centers in the United States. Equipment brokers and energy analysts say that well-maintained second-hand turbines sell for about $1.5 million to $1.7 million per megawatt, while new ones cost about $2 million to $2.5 million per megawatt.The International Energy Agency expects global data center electricity consumption to increase from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030, with electricity consumption for AI facilities expected to double during the same period. The U.S. Environmental Protection Agency states that stationary turbines and engines used by data centers must comply with federal air pollution standards. Finnish manufacturer Wärtsilä told Reuters that it has received orders for about 3 gigawatts of gas engine generators from U.S. data centers since spring 2025.

first_img Amazon expands AI cabinet delivery, World Semiconductor, Wistron, and Hon Hai show strong shipments

The supply chain reports that Amazon, a leading North American cloud service provider under AWS, has recently expanded the delivery of AI server cabinets, with strong shipping momentum from partners such as TSMC, World Advanced, Wistron, and Hon Hai. Amazon has recently revised its capital expenditure for this year, increasing the original estimate from $200 billion to $220 billion, primarily for AI infrastructure, including data centers, servers, and network equipment.Amazon CEO Andy Jassy is optimistic that the demand for AI computing power will continue to exceed supply until 2027, and he has even seen strong demand for 2028. AWS's AI business and self-developed chip scale have grown to $25 billion, maintaining triple-digit percentage growth. Annapurna Labs' customized ASIC chips are mostly produced by TSMC using advanced 3-nanometer processes.World Advanced is an important partner for AWS's self-developed AI chips and is optimistic that its quarterly revenue and profits will reach new highs this season, with continued growth expected in the fourth quarter. The demand for 3-nanometer AI accelerators is better than previously assessed, and designs can be finalized before the end of the year. AWS's ASIC AI cabinets are assembled by Wistron, which also supplies switch trays, while Hon Hai is involved in CPU trays. Hon Hai's rotating CEO, Liu Young-way, stated that this year's ASIC project has made new progress, setting a target for ASIC server market share at over 40%.

hot_img MLCC supply is in short supply: customers are raising prices two to three times to secure materials, with delivery times extended to 12-16 months

Driven by strong demand for AI, the MLCC market has seen a surge in price chasing and material grabbing. Industry sources indicate that some customers are willing to pay two to three times the price to secure supplies from major manufacturers like Yageo and Murata, creating a situation where the highest bidder wins. Yageo has admitted that AI-related customers are increasingly looking to lock in capacity in advance to reduce supply risks. Currently, major MLCC manufacturers are fully booked, with capacity nearing full load, and delivery times have extended to 12 to 16 months.In terms of capacity, Yageo expects the utilization rate for standard products this quarter to increase from about 80% in Q2 to over 90%, while special products will maintain a high level above 90%, overall trending towards full capacity. The company is also expanding production, with new capacity set to come online each quarter. Japanese manufacturers are also optimistic, with Murata significantly raising its annual profit forecast, and Taiyo Yuden stating that demand for AI servers has exceeded expectations. Industry insiders point out that large customers typically have a higher priority for capacity allocation, while small and medium-sized customers can only compete for limited supplies by offering higher prices. Yageo has noted that more and more customers wish to secure passive component supplies for the next six months to several years through long-term contracts to mitigate supply chain risks.
app_icon
ChainCatcher Building the Web3 world with innovations.