Published
-
2026-05-23 Chips, AI infrastructure, and geopolitics
2026-05-23 Chips, AI infrastructure, and geopolitics
The situation in the Middle East and uncertainties regarding policies toward China and semiconductors are simultaneously shaking decisions on oil, stocks, and AI investments. On the other hand, announcements from Google, OpenAI, NVIDIA, and ASML showed that the AI competition is expanding from “model performance” to “deployment mode, power, and supply network.”
Politics
Semiconductor tariffs still a long way off
An article reporting USTR's statement that while there will be no immediate imposition of semiconductor tariffs, it is important to protect the supply chain.
What happened: USTR’s Jamison Greer said while it’s not time to immediately impose semiconductor tariffs, he emphasized that supply chain protection is important. The main battlefields of friction with China span both trade and manufacturing infrastructure.
Why it matters: Semiconductors are a wide range of inputs, including those used in AI servers, data centers, and automobiles, so the presence or absence of tariffs can greatly affect capital investment and procurement costs.
What to watch: The tariff story should be read through sourcing shifts, reshoring signals, and rules-of-origin changes, not only short-term price moves.
US-China AI guardrails
Reports that US and Chinese authorities are discussing safety measures and operational guardrails for high-performance AI models.
What happened: U.S. and Chinese officials are reportedly discussing AI safeguards and operational guardrails through talks in Beijing. While there is full-scale confrontation, AI risk management is the only area where dialogue remains.
Why it matters: Delineating model exports, evaluation standards, and restrictions on use will likely be the focus of future regulatory negotiations.
What to watch: For China-related AI, watch whether export controls, data-transfer rules, model-use terms, and cloud access are tightened as one package.
Immigration justice and NYC raids
Primary information from the U.S. Department of Justice announcing the appointment of immigration judges and temporary immigration judges.
What happened: The U.S. Department of Justice appointed 77 new immigration judges and five temporary judges, at the same time New York City reported a significant increase in federal immigration enforcement. Strengthening of judicial processing and strengthening of enforcement are progressing at the same time.
Why it matters: Increasing both procedural throughput and enforcement pressure on the ground will increase employment, training, and municipal administration costs.
What to watch: Immigration enforcement will matter most if it changes employment verification, status-review procedures, or DOJ operating guidance.
Restarting peace in Ukraine
An article distributed by Reuters stating that President Zelenskiy is waiting for a US proposal on a new format for peace talks.
What happened: President Zelenskiy mentioned the possibility of a new peace plan from the United States, reinforcing speculation that ceasefire negotiations would be restarted. While the war situation has become fixed, the window for diplomacy has not yet closed.
Why it matters: Progress in negotiations has direct implications for sanctions, reconstruction funding and European defense spending.
What to watch: Energy, logistics, and defense exposure should be read through both battlefield developments and whether talks create credible paths to sanctions relief or escalation.
Return of Iranian nuclear materials
An article reporting President Trump's statement that the United States will eventually recover Iran's highly enriched uranium.
What happened: President Trump has said that the United States will ultimately recover Iran’s highly enriched uranium, which Iran has refused to hand over. The exchange over the handling of nuclear materials has once again become a source of tension.
Why it matters: An impasse in nuclear negotiations is likely to lead to regional conflict and sudden changes in oil prices.
What to watch: Track crude prices together with Strait of Hormuz shipping risk, maritime insurance, and tanker rates to separate market noise from real escalation.
economy
Crude oil prices flare up again
A market article reporting that crude oil prices have risen due to speculation that US-Iran talks have stalled.
What happened: Oil prices rose again and stock markets became nervous as concerns remained about the situation in Iran. Every time hopes for a ceasefire weaken, the outlook for energy and inflation wavers.
Why it matters: Even short-term news moves both CPI and corporate profits because oil spills over into transportation, power, and manufacturing costs.
What to watch: The key distinction is whether higher prices reflect demand strength or a geopolitical premium embedded in logistics, insurance, and currency costs.
Fed remains unchanged
A record of the Fed's chairman's conference that allows you to check the Fed's current understanding regarding policy interest rates.
What happened: The most recent Fed message has been to keep interest rates where they are and remain on the lookout for shocks. There is no strong feeling that there will be a rush to cut interest rates.
Why it matters: Remaining high interest rates have a broad impact on housing, capital investment, and startup financing.
What to watch: Rate expectations should be tested against loan renewals and long-cycle project economics, where higher-for-longer costs show up first.
IMF continues to face headwinds
Official IMF page for checking downside risks and growth prospects for the global economy.
What happened: The IMF says that U.S.-China dialogue and easing of tensions are good news for the global economy, but that wars in the Middle East and high energy prices remain a downside risk. There are both reasons for optimism and reasons for caution.
Why it matters: The global economy is entering a phase where it is more sensitive to changes in energy and financial conditions than to tariffs.
What to watch: Trade data should be read through volume, logistics cost, insurance premiums, and currency moves, because each can change margin even when demand holds.
Softcat makes upward revision due to AI demand
An article distributed by Reuters stating that Softcat has raised its profit forecast due to demand for AI infrastructure.
What happened: British IT agency Softcat raised its full-year profit forecast due to strong demand for AI infrastructure. Companies’ demand for introduction is primarily expressed in infrastructure and deployment rather than software.
Why it matters: The “real demand” for AI deployments tends to be in servers, networks, maintenance, and memory before SaaS contracts.
What to watch: Vendor differentiation will come less from license price than from delivery lead times, maintenance capacity, and long-term support.
NVIDIA and IREN partner for 5GW
Primary information that NVIDIA and IREN announced a strategic partnership to deploy AI infrastructure on a scale of up to 5GW.
What happened: NVIDIA announced a strategic partnership with IREN to deploy up to 5GW of AI infrastructure and up to $2.1 billion in investment capacity. AI is not only a model competition, but also a power and data center competition.
Why it matters: Securing GPUs alone is not enough; power generation, power transmission, cooling, and land constraints will influence investment decisions.
What to watch: AI infrastructure spending will be capped by power contracts, cooling, and facility constraints as much as by GPU pricing.
Technology
Google I/O will become an agent
Official blog summarizing Gemini and agent development features announced at Google I/O 2026.
What happened: At Google I/O, Gemini 3.5 Flash, Antigravity, and AI development capabilities for Android came to the fore. The competition is moving from competition in the performance of standalone models to competition in the design of applications and agents that actually work.
Why it matters: Developers need to redesign not only model selection, but also tool invocation, evaluation, and UI integration.
What to watch: AI product maturity should be judged by task decomposition, recovery from failure, and tool stability rather than the presence of a chat interface.
GPT-5.5 to API
OpenAI's official announcement about the GPT-5.5 API and ChatGPT deployment.
What happened: OpenAI rolled out GPT-5.5 to its API and ChatGPT to boost performance in coding, research, and tool usage. The update speed of the Frontier model is once again forcing the application side to follow suit.
Why it matters: Evaluations, guardrails, and prompts created based on old model assumptions can quickly deviate in performance and behavior.
What to watch: Claude 5.5 matters if production workflows show better cost, latency, and failure-mode behavior, not only stronger public benchmarks.
Codex goes on-prem
OpenAI official announcement explaining the partnership to connect Codex with Dell's AI Data Platform and AI Factory.
What happened: OpenAI is partnering with Dell to connect Codex with Dell’s AI Data Platform and AI Factory to expand into hybrid and on-premises environments. Generative AI is no longer exclusive to the cloud.
Why it matters: Companies are not just concerned with performance, but with data location, auditing, and privilege management.
What to watch: For sensitive code and internal data, the market signal is whether more options satisfy audit, permission, and data-boundary requirements.
NVIDIA breaks record
Official announcement showing NVIDIA's FY2027 first quarter financial results and data center sales trends.
What happened: NVIDIA posted $81.6 billion in revenue in Q1 FY2027, and data center revenue also hit a record high. Demand for AI factories is still being read as a story of expansion rather than slowdown.
Why it matters: As long as the chain of semiconductors, power, cloud, and capital investment continues, the AI market will be a tug-of-war between supply constraints and demand growth.
What to watch: AI-stack rollout speed will depend on supply delays, power constraints, and data-center readiness as much as on hardware price.
ASML supply tight
ASML's quarterly results that allow you to check the demand and supply status of advanced semiconductor equipment.
What happened: ASML expressed the view that the supply of advanced semiconductor equipment will be tight for the time being due to demand for AI. From the perspective of equipment manufacturers, the AI boom is still running into equipment constraints.
Why it matters: Lithography equipment is the core bottleneck for supply capacity at advanced nodes.
What to watch: Semiconductor investment should be tracked through equipment backlogs, delivery timing, and ramp schedules, not only demand forecasts.
Cross-sectional view
- Political uncertainties are concentrated on semiconductor tariffs, US-China AI guardrails, and the situation in Iran, and are directly affecting crude oil, procurement costs, and AI investment decisions.
- The primary battleground for AI investment is shifting from model performance to power, data centers, and manufacturing equipment. NVIDIA, IREN, and ASML materials point in the same direction.
- Corporate usage is no longer focused exclusively on the cloud, but is leaning toward on-premises, hybrid, and auditable operations. The development of Codex is a symbol of this.
- Immigration, trade, and security are all moving at the same time, so it is more practical to look at them through a common stress test rather than dividing legal, procurement, and finance vertically.
Uncertainty to track
- Will the actual timing and scope of semiconductor tariffs be clarified?
- Will the U.S.-China AI guardrail end up being an abstraction or will it fall into concrete operational rules?
- How far will the conflict over Iranian nuclear materials affect crude oil and shipping insurance?
- Will the Fed maintain its stance until the next meeting, given the high oil prices and geopolitical shocks?
- Will the acceleration of AI infrastructure investment cause power and facility constraints to become apparent before GPU prices?