From OpenAI's HBM Race to Key Points in Nvidia's Earnings: Major Issues Analyzed

OpenAI is treating securing high-bandwidth memory (HBM) as the key issue for its business. Beyond simply supplying chips, OpenAI aims to supply entire networking systems, and HBM's role in that strategy has become especially prominent. Analysts say that if HBM runs short, OpenAI's in-house systems could become useless, so its memory-sourcing capability will translate directly into business competitiveness.

Huawei's core proposal is to use its in-house Ascend 950-series AI accelerators. It plans to use 1,408 Ascend 950 units for training, and Ascend 950 and 910-series units for inference. If the project proceeds successfully, it is expected to be Huawei's first large-scale overseas export case for its AI accelerators.

Huawei has struggled to export overseas under US sanctions, so success in Egypt would be highly significant. Huawei has already been preparing its market expansion by pursuing preliminary projects in education and healthcare in African countries such as Kenya.

The US State Department is discussing forming a consortium of American companies to respond. Since US chip export restrictions began in 2023, a gap has opened in the supply of US-made chips in Egypt, which has created an opening for Huawei chip adoption.

He has argued that developers in China should be locked into American technology platforms. The Egypt case is an example of his warning extending to third countries. Huawei's chip performance still lags, but warnings point to the possibility of gradual improvement.

A market issuance in September is being discussed, but SoftBank's junk rating and widening credit default swap (CDS) spreads are expected to raise its borrowing costs.

SoftBank's current loan-to-value (LTV) ratio is about 13%, which is considered stable for now. However, if the AI market corrects, SoftBank's net asset value could fall. A drop in asset value would push up LTV, raising the likelihood of additional collateral demands and early repayment, so a close look at the sustainability of SoftBank's AI investment is needed.

He cites the "debasement trade" as the basis for this upside, meaning a phenomenon in which rising debt and falling real value progress at the same time. Growing geopolitical risks and continued central bank gold purchases are also supporting gold and silver prices.

Growth in the solar industry is driving a sharp rise in silver demand. Silver use in solar panels grew from 81 million ounces in 2016 to 200 million ounces in 2024. Experts forecast it could reach as much as 450 million ounces by 2030. That would mean half of the world's silver mine output in 2030 is consumed by the solar industry, suggesting that the shift to green energy will be an important driver of silver prices.

With Nvidia's second-quarter earnings report imminent, market consensus expects revenue of about $92 billion and a gross margin of 75%. Data center revenue in particular is expected to reach $85.7 billion, making it a key gauge of whether AI industry capital spending is continuing. Investors are focused on changes in dependence on hyperscalers and on CEO Jensen Huang's interpretation of the results.

Nvidia's results will serve as a stress test of AI companies' capital spending, and whether Nvidia can maintain high margins during the shift to next-generation chips will be closely examined.

This shows the market expects significant share price movement depending on Nvidia's results.