Yazılar

Nvidia to Exclude China from Financial Forecasts Amid U.S. Export Restrictions

/in /tarafından

Nvidia will stop factoring in revenue and profit from the Chinese market in its financial forecasts, CEO Jensen Huang told CNN on Thursday, citing ongoing U.S. trade restrictions on chip sales to the region. The decision comes as the U.S. maintains stringent export controls that limit Nvidia’s ability to sell its advanced chips to Chinese customers.

When asked if the ongoing trade discussions between the U.S. and China could lead to a lifting of export controls, Huang said he was not counting on any changes:

“If it happens, then it will be a great bonus. I’ve told all of our investors and shareholders that, going forward, our forecasts will not include the China market.”

Huang reiterated his criticism of U.S. chip export curbs, arguing that they are not achieving their intended policy objectives. “The goals of the export controls are not being achieved,” he said. “The goals have to be well-articulated and tested over time.”

According to D.A. Davidson analyst Gil Luria, Nvidia may face downside risks for 2026 if it remains unable to resume sales to China. Nvidia’s China business remains significant: in the first quarter, China accounted for 12.5% of the company’s total revenue, generating $4.6 billion largely from customers stockpiling the H20 chip before the restrictions took full effect.

The company estimates the export curbs cost it $2.5 billion in lost sales in Q1, with an $8 billion revenue hit projected for Q2. Nvidia is still exploring limited options for the Chinese market but acknowledged:

“Until we settle on a new product design and receive approval from the U.S. government, we are effectively foreclosed from China’s $50 billion data center market.”

Michael Ashley Schulman, CIO at Running Point Capital, said Nvidia’s move to exclude China from its forecasts simplifies its financial outlook:

“By zero-basing China, Nvidia removes a volatile variable that neither Wall Street nor the Commerce Department can reliably handicap.”

Temasek Joins Microsoft, BlackRock, MGX in Major AI Infrastructure Investment Consortium

/in /tarafından

Singapore’s state investment firm Temasek has officially joined the AI Infrastructure Partnership (AIP), a major global consortium backed by Microsoft, BlackRock, and MGX, according to presentation slides shared during BlackRock’s investor day on Thursday. The consortium also includes BlackRock’s Global Infrastructure Partners.

AIP, formed in September, aims to invest over $30 billion initially into building the data centers and energy facilities required to support artificial intelligence applications, such as ChatGPT. The partnership ultimately seeks to mobilize up to $100 billion, including debt financing, with a primary focus on projects within the United States.

Temasek’s involvement comes shortly after the Kuwait Investment Authority became the first non-founding sovereign wealth fund to join the group earlier this month. The consortium’s partners also include key AI players like Nvidia and Elon Musk’s xAI.

Ravi Lambah, Temasek’s head of strategic initiatives, commented on the development: “Temasek’s investment in the AI Infrastructure Partnership reflects our focus on the big shifts and trends of the future. AI is potentially the most transformative and impactful technology for all sectors and businesses.”

The financial terms of Temasek’s investment were not disclosed. As of March 31, 2024, Temasek reported a net portfolio value of S$389 billion (approximately $304 billion), according to its official website.

Exclusive: Crusoe’s ‘Neocloud’ to Buy $400 Million in AMD AI Chips for Data Centers

/in /tarafından

Crusoe, an artificial intelligence-focused cloud computing startup, revealed plans to purchase approximately $400 million worth of AI chips from Advanced Micro Devices (AMD) to power its AI data centers. CEO Chase Lochmiller told Reuters that Crusoe intends to acquire around 13,000 AMD MI355X chips for a new data center cluster in the U.S., which is expected to become operational this fall.

The data center will employ liquid cooling technology and be designed specifically to house AI chips, offering higher performance compared to older infrastructure. Crusoe will rent access to this facility, which can be partitioned among multiple clients or used entirely by a single customer.

Lochmiller emphasized Crusoe’s agility as a smaller startup, competing with larger hyperscalers by leveraging speed, nimbleness, and concentrated engineering talent.

AMD’s MI355X chips, featuring high-bandwidth memory, are optimized for AI inference tasks, providing an alternative to Nvidia’s dominant hardware in the AI chip market. While many AI cloud services rely on Nvidia chips, AMD is positioning itself to capture a share by partnering with companies like Crusoe.

Lochmiller described this approach as a validation of the “neocloud” strategy — specialized cloud infrastructure platforms tailored for AI workloads that add significant value to the AI ecosystem by supporting large-scale users.