What Jensen Huang Revealed About Nvidia’s Bold “Sun Strategy” in the BG2 Interview

AI Market Trends

Nvidia’s macro strategy rests on the premise that compute resources are moving from CPU‑centric cost centers to GPU‑centric production centers that generate massive economic value. The demand for AI compute follows a "double exponential" model, combining rapid user‑base expansion and escalating per‑query compute needs.

1.1 Computing Paradigm Shift: From Cost Center to AI Factory

Since last year’s GTC, Huang has re‑defined data centers as "AI Factories" that produce a new high‑value commodity quantified as Tokens. This reframes customer decisions from "how much does the system cost?" to "what economic return will the Tokens deliver?", moving evaluation from TCO to ROI.

1.2 The Double‑Exponential Growth Engine

The first exponential factor is the explosive growth of AI‑enabled applications and users (e.g., ChatGPT surpassing 1.5 billion MAU in under three years). The second factor is the surge in per‑query compute, driven by multi‑step, "thinking" models such as OpenAI o1, with an estimated >50 % CAGR in inference compute from 2022‑2025.

These two curves create a self‑reinforcing loop: stronger Nvidia platforms enable larger models, which boost user experience, attracting more users and further increasing inference demand.

Why Nvidia Is the Biggest Beneficiary

Decades of investment across technology, ecosystem, and supply chain have built deep moats.

2.1 Technical Moat

Silicon: High‑performance GPUs (e.g., Blackwell) tightly coupled with Grace CPUs.

Interconnect: Proprietary NVLink for high‑speed chip‑to‑chip communication.

Networking: Spectrum‑X Ethernet and InfiniBand solutions for large‑scale clusters.

System: Integrated DGX SuperPOD and NVL72 rack‑level systems.

Software: CUDA, cuDNN, TensorRT, Triton, NeMo and other libraries forming a full‑stack development environment.

Huang calls this "Extreme Co‑design", ensuring seamless collaboration across all layers.

2.2 Ecosystem Moat

CUDA has become the de‑facto AI development standard, with over 4 million registered developers and >90 % market share. Complementary tools (cuDNN, TensorRT, Triton, NeMo) lock developers into Nvidia’s workflow, creating high conversion costs for any migration.

2.3 Supply‑Chain Moat

Nvidia can place multi‑hundred‑billion‑dollar wafer and HBM orders with TSMC and SK Hynix before customer commitments, leveraging its capital strength to lock advanced packaging capacity and prioritize delivery, effectively shaping the global semiconductor supply chain.

How Nvidia Turns Moats into Dominance

3.1 Re‑shaping Value Metrics: From TCO to TVO

Huang argues that even free ASICs would lose to Nvidia because the metric shifts to "Tokens per Watt" and "Tokens per Dollar"—the economic output per energy or investment—rather than raw FLOPS or cost.

3.2 Binding "Super Buyers": Massive Investment in OpenAI

Nvidia plans to invest trillions of dollars with OpenAI over the next years, cementing its platform as the industry standard and creating a strategic feedback loop that pressures rivals to match Nvidia‑powered compute.

3.3 Platform‑as‑Product: From Component Supplier to AI Factory Designer

Through DGX SuperPOD and NVL72, Nvidia offers turn‑key, performance‑optimized solutions that simplify AI‑Factory deployment, turning a multi‑choice hardware selection problem into a single‑choice platform decision.

Current Strategic Moves (The "Sun Strategy")

4.1 Sovereign AI

Huang promotes "Sovereign AI"—national AI infrastructure built on Nvidia technology—turning geopolitical trends into a multi‑trillion‑dollar market.

4.2 Open‑Link Strategy

Nvidia’s partnership with Intel aims to open NVLink as an industry standard, displacing PCIe and creating a unified high‑performance interconnect across x86 ecosystems.

4.3 GPU Versus ASIC

Huang stresses that the rapid evolution of AI models makes ASICs a risky bet, while Nvidia’s programmable GPUs and extensive CUDA ecosystem provide the flexibility needed for the "Cambrian explosion" of AI innovation.

Long‑Term Outlook (5‑10 Years)

Nvidia envisions the AI Factory as the "GE of the intelligent economy", supplying compute as a utility and eventually extending to "Physical AI"—robots, autonomous vehicles, and drones powered by Jetson, Drive, and Omniverse platforms.

References: Steven Fiorillo (Seeking Alpha), Citi Research, Sequoia, GTC 2025 presentations.