cost

On September 3 local time, Tesla held a Cybercab launch event in Austin, Texas, USA. This vehicle is a mass-produced model designed for autonomous driving scenarios, eliminating the steering wheel, pedals, and traditional rearview mirrors, and adopting a two-seat layout, with driving entirely controlled by Tesla's autonomous driving software. The Cybercab was unveiled as a concept car at the "We, Robot" event in October 2024, with the first mass-produced vehicle rolling off the line at the Texas Gigafactory in February this year, and mass production starting in April, taking about 18 months from concept to mass production.The vehicle relies on 8 high-definition cameras to perceive the environment and makes driving decisions through an end-to-end neural network, without relying on lidar and high-precision maps. According to Tesla's data, as of September 2026, the global fleet's cumulative assisted driving mileage has exceeded 22.5 billion kilometers. In terms of energy efficiency, it is expected to travel 9.8 kilometers per kilowatt-hour, with an energy consumption of about 10.2 kilowatt-hours per 100 kilometers. The curb weight is 1412 kilograms, equipped with a 47.6 kilowatt-hour battery pack and a 163-kilowatt motor, with a laboratory range of 673 kilometers. After scaling up ride-hailing operations, the cost per mile can be reduced to $0.2, equivalent to about 0.84 yuan per kilometer.The Cybercab will be showcased in multiple cities such as Beijing and Shanghai starting in mid-September. This exhibition does not involve the sale of the vehicle in China, nor does it represent that Tesla will conduct autonomous ride-hailing commercial operations domestically.

According to the Guangzhou Daily New Flower City report, a man in Guangzhou lent 1 virtual currency to a friend, agreeing to return it two days later. After the other party failed to return it on time, he was sued, with the plaintiff requesting the return of the virtual currency or compensation at market price.The Guangzhou Intermediate People's Court ruled that the relevant virtual currency lending agreement was invalid and did not support the return of the virtual currency or compensation at market price. Virtual currency belongs to online virtual property, and the borrower should still compensate the lender for property losses, ultimately ruling to compensate 199,600 yuan based on the cost price at which the lender purchased the virtual currency.

The SEMICON Taiwan 2026 Memory Summit took place on the 1st, where Nikhil Cherian, Senior Director of Supply Chain Infrastructure at Google Cloud under Alphabet, pointed out that with the popularity of multimodal and mixed expert architectures, AI computation has shifted from being power-limited to memory-limited, with high-performance memory accounting for over 75% of the cost of AI server hardware bill of materials. In the face of capacity, bandwidth, and power consumption bottlenecks, Google is breaking through the AI memory bottleneck through a dual-track strategy of hardware offloading for inference and training, and lossless quantization software algorithms.Google adopts an offloading strategy in hardware architecture, launching TPU 8i for low-latency inference and TPU 8t specialized for large-scale training. The TPU 8i is equipped with 288 GB of high-bandwidth memory, with SRAM capacity on the chip increased threefold to 384 MiB, placing dynamic conversation states and key-value caches on the chip itself to achieve zero chip-off latency. The TPU 8t forms a super-large computing cluster with 9600 chips, achieving a shared pool of HBM at a scale of 2 PB, eliminating chip-off data transfer bottlenecks, along with TPU Direct Storage technology.Google has developed the training-free TurboQuant lossless quantization algorithm, compressing the key-value cache of large models from 32 bits to 3 bits, reducing memory usage by six times without loss of accuracy, resulting in an eightfold acceleration in attention computation, and integrating old-generation DRAM technology to extend the lifecycle of components.

IMF President Georgieva stated at the Jackson Hole annual meeting that stablecoins and tokenization could enhance global financial liquidity, making large cross-border payments cheaper and faster; however, stablecoins could also exacerbate currency substitution, capital flows, and exchange rate volatility, while undermining capital controls and monetary sovereignty. Dollar stablecoins may expand the global network effect of the dollar and marginally reduce U.S. financing costs, but they cannot replace fiscal discipline.This year's Jackson Hole Global Central Bank Annual Meeting has revealed three clear institutional divergences: BIS leans towards "marginalization of stablecoins, centralization of tokenized deposits"; ECB leans towards "central bank currencies on-chain"; while IMF acknowledges the real efficiency of stablecoins in cross-border payments but emphasizes the risks of currency substitution and capital flows in emerging markets. This has more policy implications than simply "supporting or opposing stablecoins."

The core development team of Solana, Anza, announced that the Solana testnet has enabled the first phase of the SIMD-0437 proposal, and the account rent reduction mechanism has begun testing. This proposal includes a total of 5 functional gates, with only the first functional gate currently activated, and all adjustments have not yet been completed and deployed to the mainnet.Once all 5 phases are completed, the storage cost parameter lamports_per_byte for Solana will be reduced from 6960 to 696, resulting in an overall reduction of 90%. For example, in the case of token accounts, the deposit required to maintain an account in a rent-exempt state is expected to decrease from approximately $0.16 to $0.016, thereby lowering the costs of account creation and application deployment.

P Equity Research and SemiAnalysis researcher Nick Doyle and others discussed high bandwidth flash (HBF) in X Space. Nick stated that it is still too early to determine how much the cost premium of HBF relative to HBM can shrink; existing data mostly comes from vendor claims, such as Sandisk stating that the cost per bit is about one-eighth that of HBM. Yields, testing, and other factors will improve with scale, but structural costs such as TSV, stacking, and pSLC mode will always exist, and durability is a key unknown; if wear exceeds expectations, costs will rise.The application scenarios for HBF are narrow, targeting only AI inference, especially low batch and long context MoE models, and it is not a substitute for HBM. The actual bandwidth target is about 1.6 TB/s, which is at the HBM3E level, suitable for sequential reads to load model weights, more aligned with the capacity needs of a small number of GPUs in local or private enterprises, rather than ultra-large-scale bandwidth scenarios. Heat dissipation reliability has not yet been resolved; flash memory will degrade faster at high temperatures next to GPUs, and mitigation measures such as UCIe separation and daily refresh have yet to be validated.In terms of manufacturing, Sandisk/Kioxia has experience with 3D NAND, and SK Hynix complements HBM-style stacking capabilities, but mass production is still to be confirmed. Overall, storage is shifting towards a specialized layered market, with NAND shortages expected to continue until 2028, and HBF may further impact supply and demand.

According to Glassnode data, the spot price is around $79,000, with the first significant cost-intensive zone located between $83,000 and $86,000, where there is a supply of approximately 1.05 million BTC held by long-term holders. This portion of chips has basically not been sold during the entire round of corrections and will become a key area to test whether "patient supply will sell near the break-even point."In the lower range of $62,000 to $65,000, there are about 1.44 million BTC, of which nearly two-thirds are held by recent buyers. Glassnode states that recent buyers are providing support, while long-term chips constitute the supply wall above. The cost basis for short-term holders is around $70,000.

According to CryptoQuant analyst MorenoDV_, after Bitcoin rebounded above the new whale cost basis of around $70,000, new whales realized profits reaching a record $1.2 billion.If Bitcoin can maintain above the new whale cost basis of around $70,000, while realized profits return to normal, it will indicate that new demand is effectively absorbing market selling pressure.

Venture capitalist Tomasz Tunguz pointed out that the narrative of AI infrastructure resembles a slow relay race, with bottlenecks sequentially transmitting from GPUs to memory, CPUs, and storage, each link freezing the supply chain of the next for years and locking in higher baseline costs. At the beginning of 2023, the GPU shock caused H100 rental prices to exceed $9 per hour, and server shipments fell by 22%; subsequently, manufacturers shifted capacity to HBM, leading to an 80% quarterly increase in enterprise SSD prices and over a 60% rise in DRAM.By the end of 2025, the workload of intelligent agents will push the CPU to GPU ratio to about 1:1, with the average price of server CPUs rising by 27% year-on-year; in 2026, nearline HDD annual capacity will be sold out. The construction cost of data centers has risen to about $20 billion per gigawatt, with orders for long-cycle equipment such as transformers and turbines scheduled as far out as 2029 to 2031.Tunguz referred to this as the long whip effect in the hardware sector: years of manufacturing delays amplify downstream demand shocks upstream, and when pressure is relieved at a certain bottleneck, it will be delayed in transmitting to the next link, with transformers scheduled for delivery in 2027 to 2028, NAND wafer fabs, and turbine production lines potentially facing the risk of overcapacity.

Metaplanet CEO Simon Gerovich stated that the core logic behind their bullish outlook on Bitcoin is that AI is significantly reducing the production costs of software, content, and analytics, and value will shift towards scarce assets that cannot be infinitely increased.The supply cap of 21 million Bitcoins is guaranteed by network rules, cannot be copied or diluted, and can be settled globally 24 hours a day. The long-term experience of Japanese households regarding changes in currency purchasing power has also reinforced this judgment.

The latest research report from the Hyperliquid Policy Center (HPC) states that perpetual contracts expand hedging options and improve price discovery, with no evidence found of statistically significant harm to the benchmark futures market. The report argues that perpetual contracts are complementary to traditional futures with expiration dates, rather than zero-sum substitutes.The study utilizes the natural experiment of traditional markets being closed on weekends while perpetual markets continue trading, comparing 205 weekends of Bitcoin trading and 19 weekends of on-chain crude oil perpetual (xyz:CL) samples. The report states that expiring futures require calendar-based forced rollovers, with the cost of rolling a $10 million exposure on the Monday of April 2026 being about $950,000, while on Friday it is about $110,000; perpetual positions do not have this forced cost. The median transaction price for on-chain crude oil perpetual during non-trading hours is about $1,300, approximately one percent of the benchmark WTI median transaction price.HPC also provides an example where the crude oil weekend repricing on the week of March 6, 2026, was 15.8%, with the benchmark market completely closed; if hedged through on-chain crude oil perpetual, a $10 million position loss could be reduced from about $1.58 million to approximately $62,000 (after accounting for all costs).

According to data disclosed by intangiblecoins, the net inflow for Bitcoin spot ETFs reached $1.6 billion this week, and the week is not yet over, marking the largest single-week net inflow since the historic high point in October 2025.Related statistics also show that the average cost basis for Bitcoin ETF holders is approximately $84,156.

According to a report by Reuters, Tether has ceased its Bitcoin mining operations in Uruguay as of 2025. Documents and sources indicate that the project was ultimately abandoned due to a misunderstanding between Tether and the Uruguayan national electricity company UTE regarding the terms of power supply.Tether believed the contract stipulated a minimum power supply that could be adjusted, while UTE viewed it as a maximum quota that could not be exceeded.

In a research report released by the Bank of Italy in July 2026, a "mystery shopper" empirical investigation was conducted for the first time, tracking 200 USDC transfers across ten corridors between Italy and Argentina, Brazil, South Africa, the UAE, and Japan. The results showed that the total cost of stablecoin remittances fluctuated greatly, with a minimum of only 0.3% and a maximum close to 9%. On-chain transfers accounted for an average of only 0.4%, with the bulk of costs concentrated in fiat withdrawal and deposit stages—traditional intermediary fees such as exchange buy-sell spreads, credit card fees, and withdrawal fees were the decisive factors. Compared to traditional channels like Wise, stablecoins have a cost advantage in some corridors like Brazil to Italy, but the costs are higher in corridors like UAE to Italy, showing a high degree of "corridor specificity."In terms of speed, blockchain transfers themselves take only a few minutes, but end-to-end efficiency entirely depends on the quality of the traditional payment infrastructure in the destination country. Countries with instant payment systems, such as Brazil (PIX), Italy (TIPS), and Argentina (Transferencias 3.0), can keep the entire process under 20 minutes; whereas countries like South Africa, which rely on traditional bank transfers, see the arrival time extended to 1 to 2 business days. The report pointed out that the efficiency of stablecoin remittances is jointly determined by their own infrastructure and the surrounding traditional payment infrastructure, with both being complementary rather than substitutive. The report also analyzed the impact of global regulatory fragmentation: the EU's MiCA and the US's GENIUS Act represent mature compliance frameworks; Japan's strict "safety first" access, while lowering nominal costs, complicates processes leading users to offshore platforms; countries like India and Turkey are in a transitional regulatory phase; while countries like Egypt and Saudi Arabia, with prohibitive measures, have failed to suppress demand, instead pushing transactions into gray channels.

CryptoQuant analyst Darkfost stated that Bitcoin is trying to close above the cost price of active supply today. Active supply refers to all Bitcoins that have been traded at least once in the past seven years. By excluding inactive Bitcoins in this way, the actual price is about $70,400.Given the current market conditions, this is a more reasonable cost price. The last time Bitcoin successfully closed at this level was in early April, after which it fell below this level in early June, and sellers regained control of the market.

Analyst Alex Thorn from Galaxy Research stated on the X platform that within the price range of $62,000 to $65,000, Bitcoin's trading volume exceeds that of any other price range (excluding $0 to $1,200). As Bitcoin's price rises, this area currently constitutes a solid cost support.

OKX CEO Star stated that OKX is collaborating with Anthropic to address issues related to its corporate accounts. Earlier, Bloomberg reported that OKX was suspended due to Hong Kong employees violating Anthropic's geographical restrictions to access Claude, and has prohibited employees residing in Hong Kong or passing through China from using Anthropic's Claude AI model.Additionally, Star posted that the monthly cost for AI usage is between 6 million to 8 million dollars, and has built an internal AI development platform called Oli, which integrates multiple cutting-edge models to provide advanced AI capabilities and development tools for product managers and engineers.

Eclipse founder Neel Somani published an analysis stating that by leasing diesel generator sets, AI data centers can bypass grid access bottlenecks and complete deployment within months. Taking a 50MW inference data center as an example, about 40 leased diesel units are needed (monthly rent of approximately 2 million USD), consuming about 2.5 million gallons of diesel each month (approximately 12.5 million USD), with the cooling system costing about 1 million USD, and the total hardware investment being about 1.9 billion USD (including approximately 21,000 GPUs and supporting network equipment). Based on similar leasing agreements like Colossus 2, monthly revenue could reach approximately 180 million USD.