Nvidia’s Earnings: A Record High, a China Void, and a 50-Petaflop Promise

Nvidia shares retreated after US-China summit failed to unlock H200 AI chip sales, but weekly gains remain amid strong product pipeline and analyst bullishness ahead of earnings.

Nvidia shares scaled an all-time high on Thursday, touching €201.05, only to give back those gains the next day as a high-stakes US-China summit ended without the chip-export breakthrough investors had hoped for. The 3.56% Friday downturn left the stock at €193.90, yet the weekly tally still shows a 6% advance — a testament to the powerful countercurrents shaping the stock ahead of next week’s earnings report.

The pullback was directly linked to the outcome of the Trump-Xi meeting in Beijing, where Nvidia chief Jensen Huang traveled as part of the US delegation. Expectations for a deal that would unlock sales of the H200 artificial-intelligence chip to Chinese clients were dashed. While US authorities had already approved exports to roughly a dozen firms including Alibaba and Tencent, the Chinese government continues to pressure domestic companies to source from local rivals such as Huawei. Multiple Chinese tech companies have since canceled their orders. US Trade Representative Jamieson Greer characterised export controls as a secondary topic in the talks, insisting the purchasing decision lies entirely with China.

The financial implications are substantial. Nvidia’s current revenue forecast for the first fiscal quarter — $71.8 billion to $78.5 billion — already excludes any contribution from the H200 in China. Analysts estimate a functioning licensing agreement could unlock up to $4 billion in additional annual sales. The regulatory framework negotiated by the Trump administration would require buyers to certify non-military use and pass the semiconductors through US territory, with Washington retaining a 25% cut of proceeds.

Yet the market’s attention is trained elsewhere. Underpinning the long-term narrative is Nvidia’s product pipeline, which remains firmly on schedule. The Vera Rubin AI platform is set to enter trial manufacturing in June, with first deliveries to major cloud providers expected by July 2026. The architecture, pairing a new central processor with the Rubin graphics core, delivers 50 petaflops — more than double the performance of the current Blackwell generation. That timeline effectively puts to rest speculation about production delays.

Alongside its core graphics business, Nvidia’s networking segment has emerged as a formidable growth engine. Revenue in that division surged 142% year-over-year in the last fiscal year to hit $31 billion, representing about 15% of total company sales. The strategy of selling complete AI data-center systems rather than individual chips is viewed by analysts as a key competitive advantage over Intel and Advanced Micro Devices.

Wall Street remains overwhelmingly bullish heading into the May 20 earnings call. Cantor Fitzgerald pegged a price target of $350 last week, citing order books that stretch deep into 2028. Bank of America raised its target to $320, pointing to accelerating capital expenditure in AI infrastructure, while Wells Fargo followed with a $315 target. Both banks anchor their bull cases on the massive energy and infrastructure buildout underway at hyperscalers such as Microsoft and Amazon, which are pouring hundreds of billions of dollars into AI data centres.

The options market is bracing for a post-earnings swing of more than 7%, a signal that the binary nature of the China question — and the sheer scale of Nvidia’s growth — could produce fireworks. The quarterly report will reveal just how much of that cloud capital is flowing directly into Nvidia’s order book, even as the political front in Beijing remains frozen.