July 20 Global Market’s Key Turning Point: Opportunities and Risks from the Convergence of Geopolitics, Policy, and Industry

Before the market opened on July 20, multiple major events were concentrated and landed at once globally—Middle East geopolitical conflict continued to escalate, pushing up oil prices; China’s A-shares saw a policy “combo punch” combining “national-team funds + regulatory symposiums”; and the AI and semiconductor industry chains showed diverging patterns under a computing-power bottleneck alongside order-driven catalysts. Four major themes together are shaping the market’s current pace, and they are worth breaking down one by one.

I. Ongoing Geopolitical Escalation in the Middle East: Oil Jumps, Repricing the Safe-Haven Logic

Tensions from the Iran–U.S. military confrontation have entered the ninth night and are still escalating. Throughput through the Strait of Hormuz has already fallen to extremely low levels, and global energy-supply anxiety continues to intensify. As of early trading on July 20, Brent crude broke above $91 per barrel, while WTI crude rose above $84. In just over ten days, the rebound from the lows exceeded 25%, and the geopolitical risk premium has been lifted to $14–$15 per barrel.

The core risk of the current situation lies in conflict spillover: Israel has entered a heightened state of readiness, and Iran’s missile-launch moves toward surrounding countries mean the deadlock over Strait blockade is unlikely to be broken in the short term. With shipping routes responsible for 20% of global oil and gas transport potentially disrupted, the oil-price center of gravity has already risen on a temporary basis, and prices are unlikely to quickly fall in the near term.

For major asset classes, higher energy prices will raise inflation expectations and indirectly affect the pace of global central banks’ rate cuts; meanwhile, safe-haven sentiment will be a short-term positive for precious metals and energy-industry targets, while markets that are highly dependent on oil imports will face some cost pressure.

II. A-Shares’ Stabilizing “Combo Punch” Lands: Signal Matters More Than the Size of Funds

On the evening of July 19, two state-owned central enterprises announced share buybacks in tandem: China Guoxin used special development loans and supporting funds totaling more than 50 billion yuan to increase holdings of central SOE stocks, while China Chengtong cumulatively bought nearly 13k yuan. After both clearly stated they will continue to step up further, they pledged to fully maintain stable market operations. At the same time, on July 20, the CSRC held a symposium between institutions and listed companies to hear views related to stabilizing the market.

In terms of size, the nearly 60 billion yuan of incremental funds is not “massive” in absolute terms, but this combination of “funds backstopping + regulatory messaging” sends very clear policy signals: it is a direct hedge against pessimistic market expectations, aimed at preventing panic sentiment from spreading and giving the market a window to repair.

Unlike past buybacks, this time the focus is explicitly on two directions—state-owned central SOEs and technology growth—plus earlier ongoing large net inflows into broad-based ETF products, increasing the precision of policy support. In the short term, the market is likely to enter a sentiment-repair window, but the sustainability of any rebound still depends on trading volume and alignment with conditions overseas.

III. AI and Semiconductors: Model Retreat, Computing Power Takes the Baton—Valuation Rebuilds

In recent days, the AI sector has shown clear differentiation. On one side, pricing logic for large-model companies has been questioned; on the other, demand for the computing-power industry chain has continued to be validated.

Kimi K3 launched and, just 48 hours later, was forced to pause subscriptions for new C-end users due to computing-power overload, while focusing on ensuring existing users’ experience. The key takeaway from this is: the capability gap among open-source large models is narrowing rapidly, but the real moat has long shifted away from model algorithms to computing-power infrastructure and supply-chain capabilities. With users able to switch models at zero cost, the moat of any single model company is extremely shallow; over the long run, the beneficiaries remain the upstream providers—computing-power chips, servers, and storage manufacturers.

Meanwhile, after the Philadelphia Semiconductor SOX Index saw a partial pullback and South Korea’s storage sector experienced sharp volatility before stabilizing, it has provided a window for valuation mismatches to be corrected. From fundamentals, the performance of leaders such as TSMC and ASML has continued to exceed expectations, AI compute demand is still growing rapidly, and the price-hike cycle for memory chips has not ended. When stock prices diverge from fundamentals, it often becomes an opportunity for medium- to long-term positioning.

IV. New Industrial Variables: SpaceX’s $100 billion Order Reconstructs the AI Server Supply Chain

Another major industry update on July 20: Hon Hai (Foxconn) for the first time secured SpaceX’s AI server contract manufacturing order—corresponding to 13k NVIDIA GB300 server rack enclosures, with a total value of about $52 billion, and deliveries are expected to begin starting in the fourth quarter of this year.

The significance of this order is not only scale, but also breaking the existing pattern: previously, SpaceX’s server orders were long dominated by Dell and Supermicro. Hon Hai’s entry means Taiwanese contract manufacturers are accelerating their share of the global AI server market. According to the plan, Hon Hai’s global AI server market share could exceed 40% by the end of this year, and the supply-chain focus for AI hardware continues to shift toward Asia.

Looking deeper, SpaceX building large-scale computing-power centers on its own indicates that AI compute is expanding from internet cloud vendors into real-world industries such as aerospace and automotive. The era of “physical AI” is being ushered in. As the boundary of computing demand keeps expanding, upstream supporting industries—chips, servers, liquid cooling, and other related supply chains—will face a longer favorable cycle.