Skip to main content

NVIDIA’s Incredible Run Since OpenAI's ChatGPT (Nov 2022 → Oct 30, 2025)

Creatix / October 31, 2025


In just 35 months, NVIDIA vaulted from a roughly $400–$500B company to the first in history to reach a $5 trillion market value—a milestone achieved October 29–30, 2025, powered by insatiable demand for AI compute that began with ChatGPT’s breakout in late 2022 and accelerated through Hopper (H100/H200) to Blackwell (B200/GB200/GB300). (Reuters)

As dazzling as NVIDIA’s rise has been, there’s a chill in the air this Halloween season—a reminder that every bubble eventually tests gravity. The AI boom that turned silicon into gold and lifted markets to euphoric heights could, if it bursts, unleash a correction far more frightening than any Halloween horror. When the spell of infinite growth breaks, investors and companies alike may wake to realize that not every model prints money, not every GPU pays for itself, and that the world built on artificial intelligence still runs on very real economics.

In the meantime, it's time to trick or treat, showcasing NVIDIA's incredible and historic rise. 

A quick timeline of the meteoric rise

  • Nov 30, 2022 — ChatGPT launches. A mass-market interface to large language models ignites the current AI boom. Early deployments relied on NVIDIA A100-class GPUs running on Microsoft Azure’s NVIDIA-built supercomputers for OpenAI. (Wikipedia)

  • 2023 — Hopper ramps. NVIDIA’s H100 (Hopper) moves into broad cloud preview/GA across Azure, Oracle, CoreWeave and others, becoming the default AI training/inference accelerator. Supply immediately tightens. (NVIDIA Investor Relations)

  • Mar 18, 2024 — Blackwell unveiled. NVIDIA announces Blackwell (B200) and the GB200 Grace-Blackwell superchip and NVL72 rack-scale systems, promising huge training & inference gains and a single-GPU domain across 72 GPUs via NVLink 5. (NVIDIA Newsroom)

  • June 10, 2024 — 10-for-1 stock split (after May 22 announcement), broadening ownership and index eligibility. (NVIDIA Newsroom)

  • Nov 2024 — Added to the Dow Jones Industrial Average, replacing Intel—a symbolic passing of the semiconductor torch to AI. (AP News)

  • FY2025 (year ended Jan 26, 2025) — Blowout results. Revenue $130.5B, +114% y/y, with datacenter the growth engine. (NVIDIA Newsroom)

  • Oct 29–30, 2025 — $5 trillion market cap. NVIDIA becomes the world’s first $5T company as AI infrastructure spend keeps re-rating the stock. (Reuters)

What actually happened under the hood

1) A demand shock for GPUs—kicked off by ChatGPT

ChatGPT’s viral launch immediately translated into orders for GPU clusters to train larger models and serve billions of queries. Early work ran on A100s; by Q1–Q2 2023, customers were pivoting to H100s, which delivered far better performance per watt and per dollar for transformer workloads. (Wikipedia)

2) Supply became the strategy: packaging bottlenecks & CoWoS

NVIDIA’s growth in 2023–2024 was limited less by demand than by TSMC’s advanced packaging (CoWoS) capacity. The bottleneck—and its gradual easing—concentrated pricing power with NVIDIA while reinforcing the company’s “sell every chip we can make” reality. (Tom's Hardware)

3) From chips to platform: the full-stack moat

NVIDIA didn’t just sell chips. It sold systems (DGX/HGX/GB200 NVL72), switching & interconnect (NVLink, NVSwitch, InfiniBand/Ethernet via Mellanox), and a software stack (CUDA, cuDNN, TensorRT-LLM, NeMo) that developers already knew. This platform approach deepened customer lock-in and let NVIDIA influence data-center designs (power delivery, liquid cooling, networking topologies). (NVIDIA)

4) The Mellanox bet paid off

NVIDIA’s 2019–2020 Mellanox acquisition gave it the high-performance InfiniBand/Ethernet fabric that binds GPUs into one giant accelerator—critical for scaling LLM training/inference and a multi-billion-dollar business in its own right. (NVIDIA Newsroom)

5) Product cadence = confidence

  • Hopper (H100/H200) carried 2023–2024. (NVIDIA Investor Relations)

  • Blackwell (B200/GB200), revealed March 2024, positioned NVIDIA for the next wave (agentic AI, longer-context inference, trillion-parameter real-time models). The NVL72 turns 72 GPUs into a single, huge NVLink domain. (NVIDIA Newsroom)
    This cadence created credible forward visibility that Wall Street could model into multi-year AI capex cycles.

6) The “AI Funding Loop” network effects

Clouds and specialized providers (Azure, Oracle, CoreWeave, others) buy NVIDIA GPUs; model companies (OpenAI and many more) rent that compute; users and enterprises then spend more on AI services—feeding the next capacity order. This loop kept GPU orders and backlog strong. (NVIDIA Developer)

Financial proof points that re-rated the stock

  • Exploding revenue & margins: NVIDIA posted $130.5B in FY2025 revenue (+114% y/y), reflecting unprecedented datacenter growth and Blackwell’s early ramp. Subsequent 2025 quarters kept setting records. (NVIDIA Newsroom)

  • Investor access & symbolism: The 10-for-1 split (effective June 10, 2024) broadened participation and stoked Dow inclusion chatter—then reality in Nov 2024, placing NVIDIA alongside America’s “industrial” icons. (NVIDIA Newsroom)

  • Market milestone: On Oct 29, 2025, NVIDIA became the first $5T company, capping a 12x+ move since the ChatGPT moment and underscoring AI compute’s centrality to the economy. (Reuters)

Strategy in three moves

  1. Own performance leadership (Hopper → Blackwell) and the system (NVLink, NVSwitch, networking). That made scale-out clusters efficient and sticky. (NVIDIA Newsroom)

  2. Own the developer mindshare (CUDA & friends). Rewriting years of CUDA-native code for alternative stacks is costly—creating real switching friction. (Nasdaq)

  3. Coordinate the ecosystem (clouds, OEMs, ISVs, integrators). This enabled rapid productization—customers could buy a rack-scale AI computer, not just a chip. (NVIDIA Investor Relations)

Headwinds that didn’t stop the ascent (so far)

  • Export controls to China. Successive U.S. rules in 2022–2024–2025 restricted shipments of A100/H100/H800/A800 variants and later H20, forcing NVIDIA to re-bin products and re-route demand to other geographies. The company publicly criticized the policy’s impact on U.S. firms. (Reuters)

  • Supply constraints. CoWoS packaging limits at TSMC stretched lead times; NVIDIA navigated by prioritizing key customers and rolling out more rack-scale, liquid-cooled designs. (Tom's Hardware)

  • Rivals & alternatives. AMD’s MI accelerators, custom silicon (Google TPUs, etc.), and new entrants gained mindshare. Even OpenAI explored Google TPUs to diversify capacity—but NVIDIA’s platform remained the default for most AI builders. (Reuters)

Why $5T was possible

Narrative meets numbers. The story (“AI is the new industrial stack”) aligned with hard data: back-to-back record quarters, a visible product roadmap, and an ecosystem standard. The result: multiple expansion and earnings growth—rarely this extreme at mega-cap scale. (NVIDIA Newsroom)

From chip to infrastructure. NVIDIA’s shift from selling parts to selling AI data centers as a product (GB200 NVL72 and successors) reframed its TAM from “GPU” to “the AI factory,” capturing silicon, networking, systems, and software value. (NVIDIA)

Developer lock-in. CUDA and the surrounding libraries/tooling are a decade-plus head start; the cost to switch is measured in time, money, and risk—advantages that compound at hyperscale. (Nasdaq)

The road ahead (as of Oct 30, 2025)

  • Blackwell deployments will dominate 2026 planning cycles; liquid cooling, power density, and NVLink domains become baseline for state-of-the-art clusters. (NVIDIA)

  • Policy remains a swing factor (export regimes, antitrust, industrial strategy), but the secular AI capex cycle—training and increasingly inference—continues to underwrite demand. (Reuters)

  • Competition intensifies, yet NVIDIA’s pace, platform, and ecosystem give it the pole position heading into the next model wave.

Bottom line

From ChatGPT’s viral spark to Blackwell-era AI factories, NVIDIA turned a demand shock into a platform monopoly of performance, systems, and software—and re-rated itself all the way to $5 trillion. The core lesson: in platform shifts, the company that owns the compute bottleneck and the developer stack doesn’t just ride the wave—it becomes the wave. (Reuters)

Comments

Post a Comment

Popular posts from this blog

When will the Tesla bubble burst?

December 11, 2024 When will the Tesla bubble burst?  We don't know Fools rush in. It's impossible to know exactly when the Tesla bubble will finally burst. Unfortunately for us at Creatix, we began shorting Tesla too soon. We are down almost 40% on our position as of today. We are not fooling ourselves thinking that we were ever make money on the short position. We truly doubt that Tesla can go down 40% any time soon.  We would love to add to the short position, but it would exceed our $3,000 limit on the stupid bets that we do for fun. We're not Mr. Beast. We have a very limited budget for ridiculousness. We would love to short Tesla tomorrow morning at the ridiculous share price of $424. Tesla is trading at an incredible 116 times earnings, which gives Tesla a market capitalization of $1.32 Trillion. Elon Musk added today $13.4 billion to his fortune. Yes, $13 billion in one day. Yesterday, he had added $11 billion. Yes, that's $24 billion in 2 days.  Six months ago, ...
Post a Comment
Read more

Will Tariffs Reduce the National Debt?

Creatix / June 30, 2025 The U.S. national debt has surpassed $34.7 trillion , and the cost of servicing that debt— just the interest payments—has soared to over $1 trillion annually as of mid-2025. This marks a historic shift: we now spend more just paying interest on the National debt than on defense, Medicare, or any single discretionary program. Economists warn that unless fiscal policy changes, interest costs will crowd out critical investments in infrastructure, education, and innovation, deepening the structural debt burden for future generations. From Osama to MAGA OBBA: the path to U.S. bankruptcy. Osama Bin Laden "succeeded" in putting us in a path to bankruptcy. The U.S. national debt began to increase dramatically after 9/11, marking a sharp departure from the budget surpluses of the late 1990s. In response to the terrorist attacks, the U.S. launched costly wars in Afghanistan and Iraq, while also implementing sweeping tax cuts under the Bush administration. These...
Post a Comment
Read more

How TikTok can Artificially Spread Socialism in America?

Creatix / June 29, 2025 TikTok's Socialist Movement in New York City  In one of the most unexpected political turns in recent New York history, Zohran Mamdani , the democratic socialist Assemblymember from Queens, has defeated former Governor Andrew Cuomo in the Democratic primary for New York City mayor. While the general election remains to be decided in November of this year, Mamdani is now the clear frontrunner. His socialist victory signals not just a generational shift, but the rise of a new kind of political power: one fueled by TikTok , a Chinese-owned social media platform that has become Gen Z’s ideological training ground. From Astoria to Citywide Dominance Mamdani first rose to prominence as a bold and principled advocate for tenants’ rights, public transportation reform, and wealth redistribution in the State Assembly. But his stunning mayoral primary win wasn’t just about policy—it was about algorithmic delivery powered by Chinese media company. Mamdani didn’t r...
Post a Comment
Read more