

The recent correction in A-share AI stocks represents a healthy deleveraging rather than a structural bubble burst, with IT sector margin trading already falling from a 12 percent peak to 8 percent to 9 percent of total turnover, according to JP Morgan. "The deleveraging process is largely complete, and the fundamental thesis for China's AI ecosystem remains intact," Zhang Xiaoning, China equity strategy analyst at JP Morgan, said in a July 15 report. Semiconductor ETF inflows turned positive in early July, averaging CNY 1.2 billion daily from July 1 to July 13, while tech hardware ETFs saw daily net inflows of CNY 350 million. Chinese cloud hyperscalers — Tencent, Alibaba and Baidu — carry debt ratios of about 40 percent, far below the 100 percent average that Chinese developers reached during their 2021-2022 peak and the 230 percent leverage that U.S. telecoms Global Crossing and Level 3 Communications carried during the 2001 dot-com bust, the report said. The report pushes back against "bubble burst" narratives at a time when global investors are reassessing China's AI opportunity. JP Morgan maintained its 2026 year-end targets of 100 points for the MSCI China Index and 5,200 points for the CSI 300, and said it expects China's AI ecosystem to reclaim market leadership during the August earnings season. **Hardware Bottlenecks Extend the Investment Horizon** JP Morgan identified global AI hardware supply constraints as a structural support for continued capital spending. TSMC's Arizona Fab 2 will not begin 3-nanometer mass production until the second half of 2027, while Micron is ramping HBM memory output at facilities in the U.S. and Singapore. SK Hynix's chief executive has warned that 2027 will bring the worst memory shortage in history, with meaningful HBM supply relief unlikely before 2028, the report said. The supply bottleneck means that the next two quarters will not serve as a valid test of AI capex sustainability, according to JP Morgan. This dynamic strengthens pricing power and urgency for infrastructure investment rather than undermining it. **Balance Sheets vs. Historical Bubbles** The report compared current balance sheets of major AI players against two historical episodes of excessive leverage. Chinese cloud hyperscalers all carry investment-grade credit ratings ranging from BBB+ to A+, while their U.S. peers — Amazon, Alphabet, Microsoft and Meta — are rated between A+ and AAA. Although recent bond issuance has widened option-adjusted spreads for some companies, the overall financial structure bears no resemblance to the distressed balance sheets that preceded prior market crashes, JP Morgan said. **What Comes Next** JP Morgan expects near-term rotation from AI into non-AI sectors during July, driven by earnings catalysts in brokerages, insurance and health care. But the bank said China's AI ecosystem will reclaim leadership in August when second-quarter financial results and second-half guidance are released. The bank advised investors to hold quality AI large-caps through the current volatility. This article is for informational purposes only and does not constitute investment advice.

Amazon's custom chip business has crossed a $20 billion annual revenue run rate with triple-digit growth, backed by more than $225 billion in customer commitments for its Trainium AI accelerators. "Our custom silicon business is now one of the top three data center chip businesses in the world," Andy Jassy, chief executive officer of Amazon, said on the company's first-quarter earnings call. If the unit sold its chips externally like a traditional semiconductor vendor, the equivalent run rate would be about $50 billion, he added. Trainium2 delivers about 30% better price-performance than comparable graphics processing units and has largely sold out, Jassy said. Trainium3, which began shipping early this year, is nearly fully subscribed, while much of Trainium4 — still more than a year from broad availability — has already been reserved. OpenAI committed to roughly two gigawatts of Trainium capacity starting in 2027, and Anthropic secured up to five gigawatts across current and future chip generations. Meta Platforms has deployed tens of millions of Graviton processor cores, and Uber uses Graviton chips for ride-matching. The chip momentum sits inside an AWS business that grew 28% year over year to $37.6 billion in the first quarter, the fastest pace in 15 quarters, with a 37.7% operating margin. Jassy said Trainium will save Amazon "tens of billions of dollars of CapEx each year" and provide "several hundred basis points of operating margin advantage." At roughly 30 times earnings, Amazon's stock is up about 10% this year — a modest gain for a company whose biggest profit engine is accelerating while a hidden semiconductor franchise scales beneath it. **The Economics of Owning the Stack** Amazon's approach differs from Nvidia's model. Rather than selling chips on the open market, Amazon keeps its silicon inside AWS, where it charges customers for compute time. The company also continues buying Nvidia GPUs — it announced plans to deploy more than 1 million Nvidia chips starting in 2026 — meaning Amazon profits whether customers run workloads on Trainium or Nvidia hardware inside its cloud. The financial stakes are large. Amazon expects about $200 billion in capital expenditures across the company in 2026, and its free cash flow for the trailing 12 months fell to $1.2 billion from $25.9 billion a year earlier as AI investments ramped up. The AWS commercial backlog stood at $364 billion entering the first quarter, excluding the recent $100 billion-plus Anthropic expansion — a cushion that reduces the risk that the spending outruns demand. For investors, the chip business strengthens an already compelling case. Amazon trades at about 30 times earnings, though that multiple is flattered by a $16.8 billion pre-tax gain on its Anthropic investment. Even excluding that gain, the valuation is reasonable for a company whose cloud business is growing at its fastest pace in 15 quarters while a $20 billion chip operation with triple-digit growth and $225 billion in commitments runs inside it. Trainium does not need to beat Nvidia for Amazon shareholders to win — it just needs to keep saving the company money and locking in customers. This article is for informational purposes only and does not constitute investment advice.

Gold edged higher in early Asian trading on Thursday but remained below the psychologically important $4,000 an ounce level, as sustained central bank buying continues to provide a structural floor under prices. "Central bank buying is likely to remain a structural driver for gold demand," according to a note from UBS's chief investment office, which called for gold to hit $5,200 over the next 12 months and described recent trading levels as an opportunity for underallocated investors. Spot gold traded at $3,987.40 an ounce as of 09:30 Hong Kong time, up 0.3% on the session. The precious metal has traded in a narrow range around $4,000 since the start of the second half, leaving it down more than 5% year to date. Silver traded at $58.72 an ounce, down nearly 20% in 2026. The consolidation comes even as geopolitical tensions around the Strait of Hormuz have re-escalated, a scenario that historically would have boosted haven demand. Higher interest rate expectations have offset those tailwinds, with markets pricing in at least one Federal Reserve rate hike this year after June's consumer price index showed headline inflation at 3.5%, still above the Fed's 2% target. **Central Bank Demand Provides a Floor** Central banks have added more than 1,000 tonnes of gold annually for the past three years, according to World Gold Council data, with purchases concentrated among emerging-market central banks in China, Poland and India diversifying reserves away from the US dollar. That structural bid has helped gold hold above $3,800 even as the dollar strengthened and real yields rose. VanEck precious metals portfolio manager Imaru Casanova described the recent pullback as noise, noting gold remains up roughly 22% from a year ago when it traded near $3,300. Mining equities have gained about 46% over the same period, compared with the S&P 500's 22% advance, she said in a note. **Mining Stocks Offer Value as Gold Consolidates** Gold miners offer 12% earnings yields, the highest of any sector, and trade at the cheapest valuations relative to the S&P 500 in two decades, according to Bank of America analysts. With all-in sustaining costs for the sector estimated below $2,000 an ounce in 2026 and gold averaging about $4,700 an ounce so far this year, margins remain strong even at current prices. Consensus mean estimates compiled by VanEck project average annual gold prices of $4,700 for 2026 and 2027, and above $4,000 through 2029. The next catalyst for prices could come from the Fed's July 30-31 policy meeting, where any signal of a prolonged pause would support the case for lower real rates. This article is for informational purposes only and does not constitute investment advice.