The wars of the 21st century are no longer the clash of steel torrents. The true life-and-death struggle takes place atop the invisible peak of computing power.
Whoever controls computing power holds the throat of the era.
And the GPU—this small silicon chip once seen as the source of joy for young gamers—has now become the Sword of Damocles hanging above the global throne of power.
This sword, which once stirred the blood of countless players in virtual worlds, now sets great powers on edge in the real world.
When the United States wields its sanctions like a cudgel, attempting to choke off another nation’s technological advancement with a single executive order, the tender veil of globalization is torn away, revealing the naked reality of strategic strangulation.
This is no longer simple commercial competition—it is a “stranglehold war” in the field of technology.
From this moment on, the evolution of graphics cards is no longer merely a history of technical progress, but a brutal war story—from arcade entertainment to national destiny.
To understand the nature of this war, we must go back to the beginning, to an age of chaos and heroes, to witness the fall of kings, the rise of gamblers, and the birth of empires.
01
Chapter One: The Savage Age and the Fall of the Old King
As history rolls forward, new kings ascend while the old ones fade away.
In the savage era of personal computers, 3D graphics were a luxury for the few.
Back then, to experience realistic 3D worlds, one needed a professional workstation from Silicon Graphics (SGI).
These machines were powerful but prohibitively expensive—often tens or even hundreds of thousands of dollars—affordable only to Hollywood special effects teams or military labs. The lifelike dinosaurs in Jurassic Park were born on such pedestals.
Ordinary players hoping to glimpse the 3D world had to rely on CPUs for “software rendering,” resulting in stuttering visuals and rough textures—barely better than nothing.
That pedestal began to crumble with the sudden emergence of a company called 3dfx.
In 1996, 3dfx stormed the market with its Voodoo graphics card—like thunder out of a clear sky.
It wasn’t a complete graphics card, but a pure 3D accelerator that needed to be connected via a pass-through cable to your main graphics card, dedicated solely to 3D calculations.
Though seemingly “parasitic,” this design delivered revolutionary performance. Overnight, PC gaming changed completely.
With Voodoo’s terrifying power, 3dfx swiftly unified the wild 3D frontier.
By 1997, it had seized 85% of the market, becoming the undisputed king.
Its proprietary Glide API became the bible for game developers of that era.
To achieve the best graphics and smoothest experience, there was only one option: Glide.
The 3dfx logo became a guarantee of top-tier gaming performance.
But a king’s throne is often built on fatal pride.
3dfx became intoxicated with its Glide kingdom and deaf to outside changes.
It believed its unbeatable hardware and closed API ecosystem would dominate the world.
But it ignored a quietly rising behemoth—Microsoft.
Seeing the vast potential of the PC gaming market, Microsoft aimed to control the graphics interface and launched DirectX, an API meant to unify hardware standards.
It was a strategic masterstroke. Microsoft aimed to build a platform empire, forcing all hardware vendors to play by its rules.
When Microsoft extended an olive branch, arrogant 3dfx sneered.
It refused to embrace DirectX, clinging to Glide, attempting to resist the entire Windows ecosystem.
This was a fatal strategic blunder. History has proven time and again: to fight platforms and standards is like trying to stop a chariot with your arms.
As 3dfx basked in glory, competitors like Nvidia and ATI quickly allied with Microsoft.
With rapid iterations, DirectX’s performance and features improved, eventually catching up to Glide. Game developers abandoned Glide, leaving 3dfx isolated.
Even worse, 3dfx made a series of disastrous product decisions.
It stubbornly insisted on separate 3D/2D card designs and missed the integrated graphics wave. In an attempt to control distribution channels, it acquired its board partner STB, enraging other partners, who all switched to Nvidia—crippling 3dfx’s own supply chain.
By the time 3dfx realized its mistakes, tried to open-source Glide, and launched integrated graphics cards, it was too late.
At the end of 2000, plagued by internal and external turmoil and lawsuits, the former king collapsed and was eventually acquired by its nemesis, Nvidia—marking the end of an era.
The fall of 3dfx offers a profound lesson: on the battlefield of technology, temporary technical superiority does not guarantee victory. Strategic myopia and disregard for ecosystems can topple any giant in an instant.
02
Chapter Two: The Gambler Jensen Huang and the Rise of Nvidia
The old king fell; a new king must rise.
This new king was, at his core, a true gambler.
His name was Jensen Huang—a man in a signature leather jacket, an immigrant from Taiwan to the United States.
His story is a legend forged in high-stakes risk.
In 1993, Jensen and two friends founded Nvidia at a Denny’s restaurant booth.
Their goal was simple: deliver powerful graphics for PC games.
But the road ahead was far tougher than expected.
Nvidia’s first product, the NV1, was an ambitious “multimedia card” combining 2D/3D graphics, audio, and even gamepad ports.
It used a unique “quadratic surface” rendering technique and partnered with Sega to port Saturn games to PC.
The gamble failed miserably.
The NV1’s tech path clashed with the industry-standard “triangle rendering.” When Microsoft’s DirectX adopted triangles, the NV1 became a technical island.
Its high price and poor compatibility triggered backlash. The Sega partnership broke down, and Nvidia nearly went bankrupt.
By 1996, Huang had to lay off over half the staff. The company was drowning in despair.
At that breaking point, Huang’s gambler instincts kicked in.
He abandoned fantasy, embraced DirectX, and bet everything on a new product: the RIVA 128.
In 1997, Nvidia had cash for only six more months.
Huang skipped all physical prototyping, went straight to mass production based on simulations—a historic all-or-nothing move.
He told leadership: “I know it’ll be perfect. Because if it isn’t, we’re bankrupt. So let’s make it perfect.”
“Perfect or bankrupt”—this became Nvidia’s DNA.
Luckily, the bet paid off.
RIVA 128 had flaws but was fast, supported DirectX, and sold over a million units in four months—pulling Nvidia back from the edge.
Even more crucially, Huang handled the Sega breakup with rare business finesse.
He honestly told Sega’s CEO that continuing the contract would kill both sides. He asked to cancel the contract but still receive full payment. Miraculously, Sega agreed. That $5 million saved Nvidia’s life.
This life-or-death victory forged Nvidia’s defining traits: sharp judgment, aggressive bets, and the will to survive.
Jensen Huang and Nvidia began their bloody ascent to the throne.
03
Chapter Three: A Duel of Titans—The Philosophical War of Architecture vs. Process
With 3dfx gone, peace didn’t follow—just a ten-year war between two giants: Nvidia and Canadian firm ATI.
It was a battle of philosophies.
ATI (later acquired by AMD) believed in “process is king.”
Pragmatic and manufacturing-savvy, ATI embraced new fabrication technologies early, betting on their efficiency and performance gains.
Its strategy: fast, frequent upgrades through small steps.
Nvidia, led by Huang, believed “architecture is king.”
More aggressive in design but conservative in process nodes, Nvidia let ATI take the early risks with new manufacturing processes, then followed once they matured—focusing instead on groundbreaking architecture.
Its strategy: big, deliberate leaps.
The difference showed starkly in how they approached risk.
ATI used new nodes for small chips first; Nvidia boldly launched flagship chips on new nodes, even if yields were terrible, to push boundaries and dominate later.
A classic showdown came in 2001: GeForce 3 vs Radeon 8500.
The Radeon was cheaper, had more features (like dual display), and great value.
GeForce 3 Ti500 was slightly faster—but its killer feature was the introduction of programmable shaders.
This innovation allowed developers to write their own GPU programs—unlocking a new era of lighting and effects.
At the time, few noticed. But Nvidia had seen the future.
They realized GPUs wouldn’t just draw triangles—they would become programmable parallel computing engines.
That foresight laid the foundation for a global software empire.
04
Chapter Four: The CUDA Wall—A Strategic Empire
True dominance is never about winning a single battle, but about setting the rules so that everyone must play your game.
For Nvidia, the rule was CUDA.
Before CUDA, general-purpose GPU computing (GPGPU) was a nightmare—developers had to fake graphics tasks just to access GPU power.
In 2006, Nvidia changed everything with CUDA (Compute Unified Device Architecture).
The idea came from Stanford PhD Ian Buck, who joined Nvidia in 2004.
He wanted to let any C programmer easily program GPUs.
Huang saw its potential and invested heavily. He wasn’t just selling cards—he was building a platform.
CUDA made GPU programming accessible.
No more graphics tricks—developers could use C/C++, Fortran, or Python with simple extensions to unlock massive performance gains.
It was a revolution. Scientists, engineers, and developers flooded in.
From molecular simulation to weather prediction, password cracking to medical imaging—everything was accelerated.
Nvidia built a thriving ecosystem around CUDA—optimized libraries (cuDNN, TensorRT), dev tools, documentation, and community support.
This ecosystem became a widening moat, leaving rivals far behind.
While AMD chased hardware specs, Nvidia quietly built a “CUDA Wall”—a fortress locking in developers with high switching costs and powerful network effects.
To challenge Nvidia, it’s not enough to match hardware—you must rebuild the entire CUDA ecosystem. Nearly impossible.
05
Chapter Five: The AI Coronation—And Hidden Dangers on the Golden Throne
Opportunity favors the prepared.
In 2012, AlexNet stunned the world at ImageNet, igniting the AI boom.
Suddenly, the world needed massive compute power to train models.
There was only one supplier: Nvidia.
CUDA, refined for a decade, was tailor-made for AI.
GPU chips became strategic gold, and Nvidia’s market share in AI data centers soared above 95%—a near monopoly.
Its revenue exploded; its valuation rocketed past $1T, then $2T, then $3T—becoming one of the most valuable companies on Earth.
Jensen Huang—the once-desperate gambler—was now crowned emperor of the silicon empire.
But beneath the throne, the currents shift.
Nvidia’s biggest customers are also its biggest threats.
Google, Microsoft, Amazon—flush with cash—buy Nvidia GPUs by the truckload, but also plot rebellion.
They resent Nvidia’s “compute tax” and are investing in their own chips: Google’s TPU, Amazon’s Trainium, and more.
These giants can’t topple Nvidia yet—but their ambition casts a shadow.
Meanwhile, Chinese companies under US sanctions fight for breakthroughs of their own.
No empire rules forever.
To understand how Nvidia survived near-death and rose to global dominance, you must understand Jensen Huang.
Born in Taiwan, sent to the US at 9, mistakenly placed in a Kentucky reform school where he scrubbed toilets daily, bullied and threatened—he learned resilience.
His first job was washing dishes at Denny’s.
That suffering shaped his values.
He often tells Stanford students: “I wish you enough pain and suffering.”
Because “greatness comes from character, and character is built not by smart people, but by those who have suffered.”
This philosophy permeates Nvidia.
When near collapse, Huang warned: “We are always 30 days from bankruptcy”—instilling survival urgency into the company’s blood.
He champions intellectual honesty—reassessing decisions constantly, correcting course immediately when wrong—a lesson from NV1’s failure.
He bets on “zero-billion-dollar markets,” guided by first principles—reimagining from scratch.
That’s how he saw GPUs beyond gaming and waited a decade for AI to arrive.
Jensen’s leadership is Nvidia’s operating system.
The CUDA empire wasn’t just a smart business move—it was born of pain-hardened will.
Nvidia’s resilience, foresight, and risk appetite all come from its founder’s imprint.
From a Kentucky bathroom to the throne of global computing—this is how Jensen Huang built his empire.
Source: Internet
Disclaimer: This article is created by the original author. The content of the article represents their personal opinions. Our reposting is only for sharing and discussion purposes and does not imply our endorsement or agreement. If you have any objections, please get in touch with us through the provided channels. (This article section is adapted from “Chip Technology: The Power-Driven Architectural Revolution,” Chapter 3 GPU and Light & Shadow Computing, with some edits.)
