BofA Recommends Gold Buys and AI Bond Shorts; U.S. Treasury Auctions in Focus

The approach reflects a strategic response to the current economic conditions and market volatility. BofA explained that this approach could help manage potential risks and identify opportunities for returns.

The first is the winning yield. When the rate set at the auction is higher than the rate traded in the market just before the auction, it is described as a 'tail.' The second is the bid-to-cover ratio, which shows how many times the bid amount exceeds the offering size and serves as a gauge of demand for Treasuries.

This shows how much demand overseas investors have for U.S. Treasuries. A high indirect bidder ratio means strong buying from overseas investors, which affects the overall stability of the Treasury market and the direction of interest rates.

Issuance of investment-grade bonds tied to AI and large data centers has reached about $270 billion so far this year. That is 12 times the average issuance from 2015 to 2024. The surge reflects the massive funding needs driven by advances in AI technology.

Some AI-related companies are posting negative free cash flow (FCF) as they carry out massive capital expenditures. To offset this deterioration in cash flow, companies are raising funds by issuing AI bonds. This shows the correlation between the huge capital needs for building AI infrastructure and companies' efforts to maintain financial soundness.

UBS raised its December copper price forecast from $14,000 to $15,000 per ton. It also projected that copper could rise to $15,500 in March, June, and September of next year. The upward revision is attributed to continued supply shortages in the copper market.

Cumulative copper output through June rose only 4% from a year earlier, and June alone grew just 2%. This sluggish output is adding to the global copper supply shortage.

Copper inventories at the London Metal Exchange (LME) fell by about 190,000 tons since May, while those at the U.S. COMEX rose by about 120,000 tons. Analysts interpret this as copper flowing to the U.S. ahead of possible U.S. copper tariffs. A BlackRock analyst said the current upward trend in copper prices remains favorable for investors and that the market supply-demand balance is very tight.

Semiconductors such as GPUs and HBM are currently the main bottleneck, but forecasts suggest that continued copper shortages could become a stumbling block for expanding AI infrastructure. Because copper supply could hold back AI infrastructure, the scope of AI investment beneficiaries may broaden from semiconductors to copper-related industries.

When tech stocks fall, investment money tends to shift to traditional defensive sectors such as healthcare. According to Ed Yardeni's forecast, the S&P 500 could rise to 8,450 by year-end. The analysis holds that even if tech stocks such as semiconductors correct, funds will not leave the stock market but spread to other sectors such as healthcare, energy, and financials, keeping the overall market firm.

Nvidia's H200 AI processors have begun to be supplied to mainland China, though in limited quantities. This means the U.S. government has allowed Chinese companies to purchase H200 chips in significant volumes. However, the Chinese government requires that H200s be deployed outside mainland China, so Chinese companies are pursuing ways to buy through Hong Kong or to use Southeast Asian data centers to access computing power remotely.

These workarounds show Chinese companies' efforts to secure advanced AI processors while evading U.S. technology restrictions. At the same time, Nvidia is reportedly adjusting its supply strategy within U.S. regulatory limits to maintain its position in China, a key market. The situation reflects the complex dynamics of the global AI semiconductor market.

Wall Street consensus for Nvidia's revenue stands at $91.9 billion. In its last four earnings reports, Nvidia beat market expectations each time, yet its stock fell every time. This reflects the market's perception that beating earnings estimates has become a baseline requirement.

As a result, even if second-quarter results are strong, the market may be disappointed if third-quarter revenue guidance comes in below consensus. Investors are placing greater expectations on future growth drivers and guidance than on simple earnings beats.

China's Unitree surged more than 600% after listing on the Shanghai stock market. Chinese companies now account for 80% of humanoid robot shipments, showing their strong position in the Chinese humanoid robot market. According to a Morgan Stanley survey, the component cost per humanoid robot is estimated at about $46,000 in China versus $131,000 in the U.S.

This price competitiveness suggests China holds a clear advantage in humanoid robots. Chinese companies' low production costs could exert downward pressure on prices for U.S. firms, which is expected to significantly reshape competition in the global humanoid robot market. With technological progress, Chinese companies may further expand market share on the strength of their price competitiveness.