WuBlockchain’s WhiteLine Daily said the current AI trade is still being driven by a chain that starts with application monetization and runs through compute spending into storage demand.
Anthropic run rate tops $65 billion
According to the report, Anthropic’s annualized revenue run rate had climbed above $65 billion by the end of July. That compares with about $47 billion in May and only about $9 billion at the end of 2025. The company is also preparing more than $10 billion in credit facilities for a potential initial public offering.
WhiteLine Daily said the more important signal at this stage is not simply that models have become smarter, but that AI agents are starting to generate substantial revenue. If applications are making money, the report argued, demand for GPUs, HBM, storage and data centers has a clearer foundation.
FT says Nvidia H200 shipments have begun reaching mainland China
Citing the Financial Times, the report said small batches of Nvidia H200 chips have recently been cleared to enter mainland China. ByteDance and Tencent each received about 10,000 units over the past few weeks, and other Chinese technology companies may soon receive approval for shipments of a similar size.
WhiteLine Daily added that the U.S. had previously approved purchases of H200 chips by multiple Chinese companies, but physical deliveries had still not been completed as of May. In its reading, the market is now moving from approval to actual shipment. If volumes continue to increase, China could again become an incremental contributor to NVDA revenue expectations and GPU demand forecasts.
AI server demand lifts NAND revenue
TrendForce data cited in the report showed that combined revenue for the world’s top five NAND suppliers rose 77% quarter over quarter to $68.87 billion in the second quarter. Demand for enterprise SSDs tied to AI servers has tightened supply.
By company, Samsung posted 70.7% quarter-over-quarter growth, the SK hynix group grew 89.5%, Micron rose 99.2% and moved into third place globally, and SanDisk gained 50.7%. TrendForce expects AI server demand to remain strong in the third quarter, while manufacturers are directing more capital expenditure toward DRAM and HBM, leaving limited new NAND capacity.
The report said the storage trade is no longer only about HBM. AI inference demand is also feeding through to NAND and enterprise SSDs. It also noted that SNDK, SK hynix, MU and Samsung are among the more actively traded AI-linked stocks on exchanges.
Higher Treasury yields pressure semiconductor stocks
On Aug. 18, the U.S. 30-year Treasury yield climbed as high as 5.327%, its highest level since 2007, while the 10-year Treasury yield rose to 4.739%. The Philadelphia Semiconductor Index fell 5% that day. Nvidia dropped 2.3%, Micron fell 7%, SanDisk lost 9%, and Western Digital declined 7.4%.
WhiteLine Daily said rising oil prices and inflation concerns pushed up long-end yields, weighing on high-valuation technology shares. In its view, AI fundamentals have not suddenly deteriorated, but the cost of capital is repricing richly valued tech stocks. For semiconductor and storage names that have already posted large gains this year, the report said interest rates now matter as much as earnings and supply-demand conditions.
Main market line of the day
WhiteLine Daily summarized the day’s main line as follows: agents begin to generate revenue, hyperscalers remain willing to add capex, money flows into GPUs and data centers, and AI inference then passes demand on to HBM, DRAM and NAND. As long as that chain does not break, the report said, the AI trade is not over.