The Pragmatic Optimist

The Pragmatic Optimist

Share this post

The Pragmatic Optimist
The Pragmatic Optimist
Nvidia’s Q4: Chip Wars Are Heating Up

Nvidia’s Q4: Chip Wars Are Heating Up

Blackwell’s $11B revenue ramp & Jensen Huang's warning shots to ASICs peers.

Uttam Dey's avatar
Amrita Roy's avatar
Uttam Dey
and
Amrita Roy
Feb 27, 2025
∙ Paid
38

Share this post

The Pragmatic Optimist
The Pragmatic Optimist
Nvidia’s Q4: Chip Wars Are Heating Up
4
11
Share

«The 2-minute version»

AI Boom? More Like an AI Empire: Nvidia posted a record-breaking $39.3B in Q4 revenue, up 78% YoY, handily beating expectations (again🙄). Blackwell GPUs were the star of the show, raking in $11B despite prior design delays. It turns out when 90% of the AI industry depends on your chips, you can afford a hiccup or two.

Margins Still Strong, But Watch for a Dip: Meanwhile, gross margins remained stellar at 75% in 2024 but are expected to dip into “the low 70s” as Blackwell ramps. Nvidia is prioritizing speed over cost efficiency, betting that quick deployment will outweigh any short-term pressure.

Warning Shots Fired: Cloud giants like Google and Amazon are doubling down on custom chips, trying to reduce their Nvidia dependency. Nvidia, of course, isn’t sweating with Jensen Huang💪 reminding investors that designing a chip and actually deploying it are two different battles.

The Regulatory Wild Card: Chip export restrictions remain a lurking threat, and new tariffs could further complicate things. Nvidia played it cool on the earnings call, saying they’ll comply with whatever rules come—but compliance doesn't always equal profitability. If Washington throws another curveball, Nvidia’s growth trajectory might see some turbulence in the short-term.

Still a Buy, But Not the Screaming Deal It Was? Nvidia is projected to grow its revenues at a 38% CAGR through FY27, while share buybacks should also be supportive of its current price levels. However, with growing competition and uncertainties from the chip export restriction, will Nvidia remain the alpha? Read below to unlock our price target.


Let’s Set The Stage

As Nvidia NVDA 0.00%↑ got ready to deliver its final quarterly earnings report of 2024, there was an unfamiliar chill in the air.

Markets were getting increasingly worried about Nvidia exceeding the high bar that has been set after two years of ultra-rapid GenAI-fueled growth.

The semiconductor giant is well known for supplying possibly the fastest computing accelerator chips for some of the most innovative GenAI companies in the world, which has led to Nvidia holding ~90% share in the global market for GPUs.

But every booming market eventually matures, leading to customizations and niche markets forming. And the GPU market for GenAI workloads is no different.

Over the past many months, some of Nvidia’s biggest customers have started building their own semiconductor chips to address their own customized needs, reducing their dependency on Nvidia’s chips. Then there is also the increased focus on efficiently deploying data center capital investments, which could also possibly lead to fewer capex dollars flowing to Nvidia.

At the Pragmatic Optimist, we also highlighted how the still-unresolved threat from chip export regulations could impact Nvidia’s growth outlook moving forward. A raft of such concerns has led to a chill in the air surrounding Nvidia’s future performance.

But Nvidia’s Jensen Huang & team appeared poised on the Q4 earnings call yesterday, ready to strongly defend their market share, innovation, and legacy.

⏰ We are running a limited time offer for all new annual subscribers to lock in $80 for a lifetime. Paid subscribers will get full access to 9-12 Research posts like this one every quarter with select stock picks in the AI ecosystem. Prices will go up higher at the end of March.

Go Premium For Just $80 Forever!


Blackwell GPU Sales Cross $11B As Nvidia’s Huang Remains Defiant

Nvidia’s final earnings report of 2024 showed the company amassed record quarterly revenues of $39.3 billion, up 12% from Q3 and up 78% from a year ago. With these reports, the company beat Wall Street’s expectations of $38.1 billion by 3.1%. But as most Nvidia investors would know, it’s not the binary outcome of earnings beat that matters; it is the magnitude by which Nvidia beat consensus expectations that matters.

Nvidia’s earnings report showed that the magnitude of the company’s earnings and revenue beat was the smallest since Q4 of 2022, the same quarter when ChatGPT was released to the world. While Nvidia beat revenue expectations by just 3.1%, its adj. earnings per share came in at 89 cents, 4.7% higher than what markets were expecting of Nvidia.

Exhibit A: Nvidia’s history of beating market expectations of quarterly earnings since 2020.

For the full year, Nvidia’s consolidated revenues grew 114% to $130.5 billion, with Blackwell shipments ramping up in the final quarter of 2024. On the call to discuss earnings, Nvidia’s management noted that design issues with Blackwell that the company experienced last year cost them “a couple of months,” but despite that hiccup, Blackwell sales exceeded their internal expectations, with Nvidia recording $11 billion just from the sales of Blackwell GPUs and systems in Q4.

The noted outperformance of Nvidia’s Blackwell sales meant the company was able to sustain strong revenues for its Data Center business that grew by a respectable 142% to $116 billion in 2024.

However, investors can note that the growth pace in the Data Center business has begun to normalize from the 217% pace it reported in 2023 as can be seen below.

Exhibit B: Nvidia’s revenue by end markets annually for the past ten years.

Jensen Huang defiantly defended Nvidia’s dominant position as a market leader and innovator by explaining how Blackwell will continue to position the company as a strong beneficiary of global AI capex, despite how Nvidia’s customer base was changing their demand dynamics about their AI accelerator requirements.

The recent influx of reasoning models such as OpenAI’s o3, DeepSeek’s R1, and Grok-3 has shifted the conversation of building data centers that are more efficient and specifically designed for AI inference. Huang & Co. have picked up on these cues, as they noted on the call how “Blackwell addresses the entire AI market from pre-training, post-training, to inference across clouds, to on-premise, to enterprise.”

When specifically asked about how DeepSeek changed any demand dynamics, Huang’s opinion was a lot more optimistic. He noted that reasoning AI models such as DeepSeek’s R1 result in more inference workloads, which is accretive to overall compute demand, boosting Blackwell’s outlook.

Huang also expressed confidence in continuing to see Nvidia as a key beneficiary of data center capex by noting:

We have a fairly good line of sight of the amount of capital investment that data centers are building out towards.

Huang also added:

Going forward, data centers will dedicate most of CapEx to accelerated computing and AI. Data centers will increasingly become AI factories and every company will have either renting or self-operated [AI factories].

In the annual filing reported to the SEC, Nvidia mentioned that 3 customers directly accounted for 10% or more of Nvidia’s total revenues each. Cumulatively, these three companies, as shown below, accounted for 34% of Nvidia’s consolidated revenues.

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Amrita Roy
Market data by Intrinio
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share