Nvidia (NVDA) stock just got a major price target reset ahead of earnings, but the twist isn’t more AI hype.
KeyBanc Capital Markets John Vinh’s call is built around a factor effectively hiding in plain sight, something investors haven’t fully priced in. It’s not about demand, nor is it about China, or even the usual AI narrative.
Instead, it’s a potential shift that could change how quickly Nvidia is able to ramp up sales, while fattening up its margins in the process.
Moreover, with expectations sky-high already, this quiet revamp could be an overlooked driver, making Nvidia’s Q2 earnings print one to watch more closely.
NVIDIA gets a major bump from a top-tier analyst with one surprising twist before earnings.Image source: Morris/Bloomberg via Getty Images
John Vinh, managing director and equity research analyst at KeyBanc Capital Markets, has a keen eye on the chip space and AI-infrastructure supply chains.
Among the stocks he covers closely is Nvidia, along with other chip names, and he has become a trusted voice on the sector.
On TipRanks, he attracts a 5-star rating, with an impressive 60% success rate and an 18.9% average return per rating. This makes him one of the top-ranked tech analysts tracked on the platform.
For perspective, Dan Ives of Wedbush fame has just a 49% success rate on the platform.
Ahead of Nvidia’s Q2 earnings report on August 27, KeyBanc’s John Vinh lifted his Nvidia price target to $215 from $190, on the back of better manufacturing yields.
Specifically, Vinh cites the GB200 server rack yields nearing an incredible 85%, which means that more systems are breezing through QA and shipping faster.
That improved throughput has him raising his full-year rack shipment forecast to 30,000 units, up from 25,000.
Simply put, Nvidia’s backlog is being converted into sales at a much quicker pace, on the back of more efficient execution at the factory level.
Supply is catching up, too.
Blackwell B200 availability has surged at an eye-catching 40% in Q2, and it’s set to climb another 20% in Q3.
The rollout of Blackwell Ultra is also accelerating, potentially making up 50% of Blackwell shipments by October, resulting in a more diverse product mix.
On numbers, Vinh now sees Q2 sales at $47.1 billion (consensus estimate at $45.7 billion) but trims Q3 to $50.4 billion (consensus estimate at $53.5 billion), excluding China due to license uncertainty.
If we include China, he sees even greater upside of $2 billion to $3 billion, driven by H20 and RTX6000D (B40) shipments.
Nvidia dominates the AI hardware stack with a whopping 80% market share, covering chips, systems, networking, and sticky CUDA software.
That grip naturally shows up in its financials. Nvidia posted $130.5 billion in FY2025 sales, more than 50% from last year, with U.S. hyperscalers deploying millions of accelerators in training and running their AI models.
Blackwell is a big part of that impressive lead.
Nvidia’s new Blackwell B200/GB200 platform is a beast, packing 208 billion transistors, a fresh Transformer Engine, and NVLink “super-highways” that let 576 GPUs act like one giant brain.
Related: Top tech stock analyst revamps AI ‘buy’ list
The result is that it efficiently delivers 30-times quicker inference and much speedier AI responses, like almost instantaneous chatbot replies.
Memory is another massive leap. Blackwell uses HBM3e (High Bandwidth Memory 3e special stacked DRAM right next to a potent processor for ultra-fast data access), handling up to 288GB per GPU.
Analysts feel Blackwell could contribute 80% of Nvidia’s high-end shipments in 2025, while Nvidia is set to control north of 70% of HBM supply.
Nvidia is set to report its fiscal year 2026 Q2 results Aug. 27, and here’s what Wall Street expects:
The big swing factors?
The first is Data Center momentum, which has been massive over the past several quarters, particularly at $39.1 billion in Q1.
Second, there’s the early Blackwell ramp with more B200 chips flowing in and GB200 racks passing quality checks a lot quicker, pulling revenue forward.
Related: Billionaire fund manager doubles down on Nvidia, partner in AI stack shift
Finally, there’s the plumbing, where networking gear like NVLink switches and full racks are moving abroad with memory (HBM) and packaging bottlenecks.
Wall Street models expect Nvidia to beat Q2 numbers, where some see Q2 revenue hitting closer to $47 billion, but Q3 staying near $50 billion to $52 billion, depending on geopolitics.
Investors shouldn’t expect a clean boost from China in Nvidia’s guidance just yet.
The U.S. only started issuing H20 licenses on August 8, and reports surfaced from Beijing, where it discouraged its use for sensitive work.
Meanwhile, Nvidia is cooking up a more robust Blackwell-based “B30A,” but still needs customer testing and volume.
More News:
On top of that, Washington is weighing a 15% revenue-share tax on China’s AI-chip sales from Nvidia and AMD.
According to White House press secretary Karoline Leavitt, despite the hurdles, the Commerce Department is still focused on getting the deal done.
“The legality of it, the mechanics of it, is still being ironed out by the Department of Commerce, and I would defer you to them for any further details on how it will actually be implemented.”
Related: Legendary fund manager Stanley Druckenmiller buys $133 million of under-radar AI stock
