Major Client Unveils on Earnings Day｜Rewire News Morning Briefing

NVIDIA's largest client revealed benchmark tests of competing chips on its earnings day, and Iran lost the negotiation table the moment it lost the Strait of Hormuz.

1｜OpenAI Unveils Jalapeño Chip on NVIDIA's Earnings Day, Officially Kicking Off the Inference Battlefield

OpenAI released the first benchmark tests of its self-developed inference chip, Jalapeño, at the Hot Chips 2026 conference. According to SemiAnalysis's InferenceX tests, Jalapeño achieves 1.5 to 1.9 times the throughput of NVIDIA's GB300 per kilowatt across three major language models, with end-to-end latency reduced by 1.7 to 3.6 times. The gap is even larger under interactive loads, being 2.1 to 4.1 times faster. The power consumption is 700 watts, compared to the competitor's flagship at 1,400 watts.

The timing itself is a signal. NVIDIA is set to release its largest quarterly earnings report ever (with consensus revenue of $92 billion) after the market closes tonight, and OpenAI chose to announce its benchmarks on the same day. Jalapeño was co-developed with Broadcom, taking only 16 months from team formation to tape-out. According to Tom's Hardware, the deployment timeline is small-scale by the end of 2026, with mass production in 2027. However, by then, NVIDIA's product line will also have evolved. The real change is not who is faster, but the inference chip market shifting from "NVIDIA monopoly" to a dual-track pattern of "client self-development + NVIDIA".

(Source: SemiAnalysis / TechCrunch / Tom's Hardware)

2｜The U.S. Informs Allies to Shift from Strikes to Sanctions, All Mines Cleared from the Strait of Hormuz

According to an exclusive report by Axios, U.S. Secretary of State Rubio recently informed several foreign ministers that the U.S. will "temporarily" not initiate a new round of strikes against Iran, shifting to economic pressure and ensuring the navigation of the Strait of Hormuz. On the same day, Trump announced that the U.S. Navy had cleared all mines from the main shipping channel of the Strait.

The underlying logic of this shift is a reversal of leverage. U.S. officials told Axios that the mine clearance is a "turning point" in this war, neutralizing Iran's major leverage over the global energy market. In the past two weeks, Iran's main oil export port, Khark Island, has seen almost no tankers dock. Crude oil prices have dropped over 5% this week. A U.S. official stated, "The Iranians have lost control of the Strait, and now the U.S. controls it." This policy is expected to last at least until after the midterm elections. Pakistan's Army Chief Munir mediated in Tehran on Sunday, but the White House stated there is currently no negotiation schedule with Iran. (Continuation of yesterday's report)

(Source: Axios / ABC News / CNBC / Times of Israel)

3｜Anthropic Aiming for $2 Trillion IPO in October, Set to Become the Largest in History

According to reports from Fortune and Yahoo Finance, Anthropic's investors are pushing for an IPO in October, targeting a valuation of $2 trillion or more, surpassing SpaceX's valuation of $1.77 trillion at its June IPO, making it the largest IPO in history. Goldman Sachs, JPMorgan Chase, and Morgan Stanley are co-managing the offering, with the public S-1 filing expected to be submitted by the end of August. According to CNBC, the filing will explicitly list "AI rebound" as a risk factor.

Supporting this valuation is revenue growth. According to Bloomberg's report on August 17, Anthropic's annualized revenue has surpassed $65 billion, growing more than sevenfold since the end of 2025. Its market share in the enterprise API sector reached 32% in Q2, surpassing OpenAI's 25% for the first time. From a safety lab to the world's most expensive IPO, Anthropic has traversed this path in less than four years. Valuation is a signal of power, indicating how much market believes there is still room for AI infrastructure.

(Source: Fortune / Yahoo Finance / Bloomberg / CNBC)

4｜NVIDIA to Release $92 Billion Earnings Report Tonight, Surprise Margin Narrowed to 4.6%

NVIDIA will release its Q2 FY2027 earnings report after the market closes on Wednesday, with Wall Street consensus revenue of $92 billion, a year-on-year increase of about 96%. The company's own guidance is around $91 billion, plus or minus 2%, with consensus slightly above the median by 0.9%. According to Yahoo Finance, NVIDIA has exceeded its own guidance for 13 consecutive quarters, but the extent of the surprises has steadily narrowed from 22.8% in Q2 FY2024 to 4.6% last quarter.

The narrowing itself is a signal. It means the market has already priced in NVIDIA's inertia of exceeding expectations, and any number below expectations will be magnified in response. On the same day, OpenAI released benchmark tests of competing inference chips (see item 1), and Google reportedly spends $900 million a month renting 110,000 GPUs. NVIDIA's moat is not its quarterly revenue, but the depth of reliance on it throughout the entire AI computing infrastructure cycle. The numbers tonight will answer not "how much was earned," but "is the reliance deepening or starting to loosen." (Continuation of yesterday's report)

(Source: Investing.com / Yahoo Finance / IG International)

It is also worth noting ↓

Dwarkesh Patel published an article in July arguing that AI computing prices could surge 10 to 15 times. Top labs have seen revenues increase tenfold annually, but computing capacity has only increased threefold, making it impossible for profit margins to sustain at 95% to fill this gap. SemiAnalysis founder Dylan Patel predicted in the same podcast that Anthropic and OpenAI will control most of the available computing power globally by 2028. (Source: Dwarkesh Podcast / SemiAnalysis)

Stability AI raised $76 million, continuing to attract investment in the image generation sector amid consolidation. The new financing comes after a significant reduction in valuation during the last round, with the market betting on survivors. (Source: TechCrunch)

Perplexity is collaborating with NVIDIA to develop portable AI computing devices. The search engine company is venturing into hardware, while NVIDIA continues to expand its ecosystem on the inference side. (Source: TechCrunch)

OpenAI's data center operations head Chris Malone has left the company during a period of accelerated computing infrastructure development. Data centers are the heaviest assets for AI companies, and the departure of an operations executive during an expansion phase is noteworthy. (Source: TechCrunch)

Apple's M6 chip will become the first 2nm processor, skipping Pro and Max versions directly to the M7 series. TSMC is the first customer for the N2 process, with the base model of the MacBook Pro set for an update by the end of this year, while professional users will have to wait until 2027. (Source: MacRumors / Cult of Mac)

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.