Despite China being a larger-than-expected drag, Nvidia’s overall Q2 guidance came in line with sell-side expectations—and about $2–3 billion above buyside forecasts—for a total delta exceeding $5 billion.
The notable part is that the first half of 2025 was operationally challenging for Nvidia (NASDAQ:). The second half is when the company is expected to be firing on all cylinders. Consequently, these strong results set a floor as investors look ahead, and we expect that every dip will get bought.
Nvidia CEO Jensen Huang highlighted that since the beginning of the year, there have been several positive developments:
Thinking models like DeepSeek are game changers for incremental demand
Canceling the AI Diffusion rule
Agentic AI proliferation
Industrial AI
This is providing management with confidence in the durability of the upside.
Jensen drew parallels between the electricity build-out that began in the 19th century and the required investment in infrastructure to take AI to the next level. We have tested the waters of what’s possible with generative AI, but for AI to transform industries, there will be a new wave of significant investment.
China Dragged Results, But Only Upside From Here
China is expected to be an $8 billion drag in Q2, exceeding analyst expectations of $5 billion.
Importantly, the current guidance includes nothing for China AI chips. This means any workaround success will result in incremental revenue. Nvidia sees China as a $50 billion total addressable market (TAM) if a reasonable regulatory framework is achieved.
Jensen Huang expressed confidence that the current administration is focused on global AI leadership, but headlines about China remain mixed. As such, any China upside is more like option value than a baked-in assumption.
For context, China represented a low-to-mid teens percentage of Nvidia’s revenue in recent quarters—and more than 20% before export restrictions. This implies significant upside if the market reopens.
Set Up for the AI Value Chain Reminiscent of 2022
Investors are skeptical about the AI trade. In 2023 and early 2024, earnings blew past expectations. But the last two quarters have seen only “in-line” to slightly better results.
Consequently, stock performance has been underwhelming, with most tech names flat or down significantly in 2025.
What’s Driving the Skepticism?
Here’s the consensus view—and where we differ:
Investors think everyone is losing market share. But after Nvidia’s latest earnings, while its stock went up, the broader AI value chain was left behind. The sentiment seems to have flipped multiple times:
Late last year: Nvidia losing share to custom ASICs
Now: ASICs and networking are losing share to Nvidia
Always: AMD (NASDAQ:) is losing share to everyone
But it’s not possible for everyone to lose share in a growing market. What’s more likely is that new products are taking longer to ramp, making it difficult to deliver earnings blowouts.
- Nvidia had a gangbusters H100 intro as customers could “plug and play” these GPUs in their existing infrastructure. But the transition to Blackwell has been more challenging than expected, given that the systems, which are what everyone wants (GB200s), are more complex and often require brand-new infrastructure.
- AMD made a strong entry to the market with the Milano GPU, but it is taking the company some time to scale and find customers that are willing to do some heavy lifting rather than going with CUDA.
- ASICs, similarly, made a big splash with Broadcom (NASDAQ:) CEO stating that custom ASICs will take over the cloud market by 2030 …but now it’s time to find the customers.
We expect that Nvidia will be able to pull up the entire value chain once the Blackwell systems come through in 2H25, very similar to what happened in ’22.
In summary, fear and uncertainty are creating significant dislocations, creating opportunities to buy dips on some of the highest quality companies across the data center value chain. Gear up for 2H25!