NVDA

Q4 2025

Date: 2025-02-26Rank: #23Forward Promise: very_bullish

NVIDIA reported record Q4 revenue of $39.3 billion, up 78% year-over-year and 12% sequentially, driven by a 93% YoY surge in Data Center revenue to $35.6 billion. The company successfully delivered $11 billion in Blackwell revenue during the quarter, marking the fastest product ramp in its history, while Hopper demand remained robust. Gross margins dipped to 73.5% non-GAAP due to Blackwell ramp costs but are expected to recover to the mid-70s later in the year. Looking ahead, NVIDIA guided for Q1 revenue of $43 billion (+/- 2%), fueled by the continued scaling of Blackwell systems and networking growth.

Bullishness Score

92.27

μ Mean

97.24

σ Uncertainty

1.66

Forward Promise

8.5

Management Tone

Management exuded high confidence and urgency throughout the call, emphasizing the unprecedented speed of the Blackwell ramp and the emergence of new 'scaling laws' in AI. Jensen Huang was particularly energetic, framing the current state as just the 'beginning' of a new computing era, while Colette Kress maintained a disciplined focus on operational execution and margin recovery.

Confidence: HIGH — Management consistently reaffirmed demand strength, provided specific forward-looking product details (Blackwell Ultra, Rubin), and dismissed concerns about custom ASICs or market saturation with detailed technical rebuttals.

Strategic Signals

Management emphasized the transition from 'perception AI' to 'reasoning AI,' identifying a new 'inference time scaling' law where models require significantly more compute to generate answers. This signals a sustained demand driver beyond pre-training, positioning NVIDIA's Blackwell architecture—specifically designed for high-throughput reasoning—as the critical infrastructure for the next wave of AI models like OpenAI's O3 and DeepSeek R1.

The company highlighted the 'Three Scaling Laws' (pre-training, post-training, and inference time scaling) to argue that AI compute requirements are compounding rather than slowing. This narrative counters investor fears of a near-term pause in CapEx, suggesting that the need for inference optimization and model customization (e.g., reinforcement learning) will drive continuous upgrades.

NVIDIA is aggressively expanding its Total Addressable Market (TAM) by targeting 'Physical AI' and robotics. The introduction of the Cosmos platform and partnerships with automotive giants like Toyota and Uber indicate a strategic pivot from digital-only AI to real-world automation, potentially unlocking a multi-billion dollar revenue stream distinct from the current data center boom.

The networking segment is poised for a rebound driven by the transition to NVLink 72 and Spectrum-X Ethernet. Management noted that Spectrum-X is becoming the standard for AI factories, with major wins like Microsoft's Stargate and Cisco's integration, signaling a long-term strategic shift away from traditional InfiniBand toward high-performance Ethernet in enterprise AI.

Key Metrics

Q4 Revenue$39.3B+78% YoY, +12% QoQ

Data Center Revenue$35.6B+93% YoY, +16% QoQ

Blackwell Revenue$11BN/A (New Product)

Gaming Revenue$2.5B-11% YoY, -22% QoQ

Non-GAAP Gross Margin73.5%-200 bps QoQ

Q1 Revenue Guidance$43.0B+9.4% QoQ (midpoint)

Guidance

Q1 Revenue: $43.0 billion +/- 2%

Q1 Gross Margin: 71.0% +/- 50 bps (Non-GAAP)

FY2026 OpEx Growth: Mid-thirties percentage range

Automotive Revenue: ~$5.0B for fiscal year

Bottom Line

Price Target: N/A

NVIDIA's Q4 earnings call reinforced its position as the undisputed enabler of the AI industrial revolution. The company successfully navigated the challenging Blackwell ramp, generating $11 billion in revenue from the new architecture in its first quarter, proving that demand remains insatiable. Management's articulation of 'Three Scaling Laws'—pre-training, post-training, and inference time scaling—provides a compelling framework for sustained growth, effectively arguing that AI compute requirements will compound rather than plateau. While gross margins face temporary pressure due to system complexity, the trajectory for the rest of fiscal 2026 is clear: massive volume growth driven by the shift to reasoning AI and the physical AI expansion into robotics and automotive. The company's moat widens with its full-stack integration (hardware, networking, software), making it nearly impossible for competitors or custom ASICs to displace NVIDIA in the near term. With the stock trading on the forward promise of continued exponential growth, NVIDIA remains a core holding for any AI-focused portfolio.

Forward Promise

Overall: 8.5 · very_bullish

Unapologetically bullish: 'We're at the beginning of reasoning AI.'

Management demonstrated supreme confidence in the sustainability of the AI boom, backed by $11B in immediate Blackwell sales and a clear roadmap for the next 18 months.

Red Flags

Gross margins contracted to 73.5% in Q4 and are guided to approximately 71% in Q1, representing a notable decline from the mid-70% levels seen in previous quarters. Management attributed this to the costs associated with ramping the complex Blackwell systems, but the pressure on profitability introduces a new financial headwind that could persist if manufacturing complexities remain high.

The complexity of the Blackwell supply chain, involving 1.5 million components and 350 manufacturing plants per rack, poses a significant execution risk. While management claimed the ramp is 'fully recovered,' the sheer logistical difficulty of scaling these systems leaves little room for error, and any hiccups could delay revenue recognition or impact yields.

Geopolitical risks remain a persistent overhang, with China revenue remaining 'well below levels seen onset of export controls.' Management noted that tariffs remain an 'unknown,' creating uncertainty around cost structures and potential market share losses in one of the world's largest semiconductor markets if regulations tighten further.

Gaming revenue declined 22% sequentially due to supply constraints, highlighting that NVIDIA's ability to meet demand is not limited to just data center GPUs. This raises concerns about whether overall substrate capacity or packaging bottlenecks could constrain growth across multiple business segments simultaneously.

Macro Insights

Global CapEx: Management indicated that data centers worldwide are shifting the vast majority of their CapEx budgets toward accelerated computing and AI infrastructure, treating data centers as 'AI factories' for token generation.

Geopolitics: China revenue remains at approximately half of pre-export control levels, and new tariffs represent an 'unknown' variable that could impact costs or market access in the coming year.

Technology Adoption: AI has 'gone mainstream,' penetrating every industry from delivery services to healthcare, suggesting a broad-based, durable economic shift rather than a cyclical tech bubble.