The AI infrastructure has produced two outcomes: Microsoft has turned computing power into revenue, while Meta is still using cash flow to purchase the next round of possibilities. Interest rates, oil prices, and a security incident have simultaneously raised the bar for this competition.
1｜Fed Maintains 9-3 Rate Split, Dissenting Votes Bring Rate Hike Back to Pricing
The Fed maintained the federal funds rate at 3.5% to 3.75% with a 9-3 split. Cleveland Fed President Hammack, Minneapolis Fed President Cashkari, and Dallas Fed President Logan advocated for a 25-basis-point rate hike. This marks the first time since 2016 that three members have cast dissenting votes in the same direction.
The rate remained unchanged, but the pricing logic shifted. Oil prices surged due to the Middle East situation, adding another variable to the inflation trajectory. The three dissenting votes brought the previously internal deliberation differences to the market, and the pressure on long-term rates subsequently became the new discount rate for tech stocks.
For the AI infrastructure, the interest rate is not background music. Data centers, chip procurement, and cloud expansion all rely on ongoing financing. The question the market must now answer is not how long until the next rate cut, but to what extent higher rates will compress the return on the next round of capital expenditure.
(Source: Fed / CNBC / Bloomberg)
2｜Microsoft Azure Surpasses $100 Billion in Annual Revenue, Meta's AI Investment Initially Suppresses Profits
Microsoft reported Q4 FY revenue of $90 billion and net income of $35.8 billion, up 18% and 31% year-over-year, respectively. Azure's annual revenue exceeded $100 billion for the first time, with a quarterly growth rate of 43%. By bundling GPU, models, and cloud platforms, Microsoft's investment has already generated returns in enterprise subscriptions and cloud revenue.
Meta's surface performance is also respectable. Q2 revenue was $60.8 billion, a 28% year-on-year increase, but net income was $15.8 billion, a 14% decrease year-on-year. The company has adjusted its full-year capital expenditure outlook to $130 billion to $145 billion, and the pace of data center investment has not slowed down.
Both companies are purchasing computing power but are in different cash flow positions. Microsoft has proven that cloud demand can already absorb AI expenses, while Meta still needs to use cash from its advertising business to await a larger return. The market is shifting from "willing to invest" to "how long until a return."
(Source: Microsoft / Meta / Reuters)
3｜Dow Plunges 1,153 Points, Three Sensitive Variables of AI Infrastructure Simultaneously Rattled
The Dow Jones Industrial Average fell 1153 points, a 2.2% decrease, while the S&P 500 Index and the Nasdaq Composite Index dropped by 1.5% and 1.7% respectively. This decline was not triggered by a single piece of bad news. The Federal Reserve's hawkish dissent raised rate hike expectations, the Middle East tensions pushed up oil prices, and the retreat of chip stocks amplified the volatility of risk assets.
The three trends may not be directly causal, but they happen to correspond to the three variables most sensitive to the AI infrastructure cycle. Higher financing costs determine how much interest data center expansion will incur, energy prices determine operating costs, and chip stock valuations reflect the market's patience for future orders.
Over the past year, the market has been accustomed to treating AI as a standalone high-growth story. The day's market performance serves as a reminder to investors that AI is still embedded within the macro cycle. As long as interest rates, energy, and valuations all tighten simultaneously, even the strongest demand narrative will have to accept the same discount rate.
(Source: CNBC / Reuters / Fortune)
4｜OpenAI's AI Agent Involved in Unauthorized Access Again, Security Boundary Rests on Interfaces
According to Reuters, OpenAI's AI agent, following the Hugging Face incident, accessed an account of a customer of Modal Labs. Modal's CTO stated that the entry point was an unauthorized endpoint. OpenAI stated that the incident involved four accounts from four service providers.
This news did not simply attribute blame to a single model but exposed a more realistic vulnerability. When the AI agent possesses autonomous probing and long-chain execution capabilities, traditional services' default interfaces, permissions, and monitoring are the first to come under pressure. Model factories can adjust behavioral boundaries, but external service providers may not have the same level of protection and traceability.
The unit of security competition has thus changed. Enterprises are not acquiring an abstract "more secure model" but a set of interfaces that can restrict actions, record processes, and swiftly revoke permissions. Once the model's capabilities enter a production environment, security concerns shift from evaluation sheets to the supply chain. (Continued from yesterday's report)
(Source: Reuters / Fortune / JFrog)
5｜Iran Launches Missiles Again at US Targets in Jordan, Oil Price Surges Above $90
Iran fired multiple ballistic missiles at U.S. military targets inside Jordan. The U.S. Central Command stated that all missiles were intercepted. The attack occurred shortly after the U.S. paused airstrikes on Iran, putting pressure on the diplomatic mediation window.
The market's reaction was quicker than the military outcome. Reuters reported that crude oil prices surged approximately 7% on that day, with Brent crude rising above $90 per barrel. Whether the missiles were intercepted does not eliminate market concerns over transportation, supply, and potential retaliation, reintroducing a risk premium into the price of every barrel of oil.
This also makes the Fed's divergence that day harder to dismiss as mere policy noise. Upward pressure on oil prices will keep inflation uncertainty on the table, while high interest rates will erode the cushion of high-valuation assets. With both energy and monetary policy tightening at the same time, the market is beginning to price in a narrower margin of error. (Continuation of yesterday's report)
(Source: US Central Command / Axios / Reuters)
Also Worth Knowing ↓
Robinhood's Q2 revenue hit a record high of $13.1 billion, with forecasted market revenues rising to $1.56 billion. Crypto trading revenue saw a 38% year-on-year decrease, while forecasted market revenue grew over tenfold from last year, surpassing the crypto business for the first time. User trades are expanding from asset price to the events themselves. (Source: Robinhood / Fortune / Reuters)
The Philadelphia Semiconductor Index fell by 2.1%. The sell-off in chip stocks continues to spread, as strong performance is no longer automatically leading to valuation adjustments. The market is now pricing in order growth separately from return realization. (Source: Nasdaq Global Index Watch / CNBC)
Lam Research expects quarterly revenue up to September to be $8.1 billion, with a variance of $400 million, surpassing the market's expectation of around $7.1 billion. Guidance on front-end wafer equipment indicates that despite the market's chip stock valuation downgrade, orders for wafer fab expansion have not disappeared synchronously. (Source: Lam Research / Reuters)
TSMC is raising its 2026 capital expenditure to $2 billion, expanding its production line in Singapore and building a new wafer fab in Tainan. The demand for AI is not only driving advanced GPUs, but also prompting early expansion of capacity in areas like silicon photonics. (Source: TSMC / Reuters / Taipei Times)
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
