Skip to main content

The Rise of NVIDIA, Part 2: The Chip That Almost Killed Us

The Chip That Almost Killed Us

🎬 Prefer to watch? Here's the 60-second version:

Watch on YouTube ▶

🎮 The Rise of NVIDIA — a 20-part series. See all parts »  |  « Part 1: A Booth at Denny's

By the middle of 1995, NVIDIA had a chip, a launch partner, and a theory about the future of graphics. Two of those three things were about to turn on the company. The chip was the NV1, unveiled on May 22, 1995, and sold at retail by Diamond Multimedia as the Diamond Edge 3D. It was ambitious to a fault: a single PCI card that fused 2D and 3D graphics, a 32-channel sound engine, and even a pair of ports for Sega Saturn controllers. NVIDIA wasn't shipping a graphics card. It was shipping a whole multimedia machine on a board.

The theory underneath it was the real gamble. And within a year, that theory would nearly end the company.

A bet on curves

Everyone in 3D graphics had to answer one deceptively simple question: what is the basic unit of a curved surface? The industry's eventual answer was the triangle — flat, cheap, easy to reason about, and easy to stitch together by the millions. NVIDIA's founders bet the other way. The NV1 rendered using quadratic texture mapping: curved quadrilateral patches instead of flat triangles. In theory, a curved primitive could describe a rounded shape with far fewer pieces. A sphere that took hundreds of triangles might be built from a handful of quads.

It was elegant. It was also almost identical to how Sega's brand-new Saturn console drew its worlds. That was not a coincidence — it was the strategy. If the NV1 spoke the same geometric language as the Saturn, then Sega's library could cross the bridge to the PC with minimal effort. Saturn hits like Panzer Dragoon and Virtua Fighter Remix shipped in NV1-compatible form, bundled to prove the point. NVIDIA had effectively wired its survival to Sega's platform. Win together, or don't win at all.

The memo that stranded a chip

Then Microsoft moved. As it built Direct3D — the graphics half of DirectX that would soon define what a Windows game was — it standardized on triangles, exclusively. There was no seat at the table for quadratic patches. In a single architectural decision, the world's most important software company declared NVIDIA's entire rendering philosophy a dead end.

The effect was brutal and fast. A wave of cheap, triangle-native accelerators — the S3 ViRGE, the Matrox Mystique, the ATI Rage, the Rendition Vérité — poured into the exact market the NV1 was supposed to own, and every one of them was aimed at the standard developers were now told to build for. The NV1's clever curves became a liability. Its all-in-one design, bundling sound hardware many buyers already had, only made it pricier and harder to sell. Reviewers found the 3D credible but awkward; developers found the tooling painful; retailers found the boxes sitting on shelves.

The numbers told the story without mercy. Diamond Multimedia wrote off roughly $5 million in unsold Edge 3D inventory in 1996, and NVIDIA quietly stopped selling the NV1 in the first quarter of that year. A company of barely a hundred people had shipped its first product straight into a wall. Years later, Jensen Huang would describe the NV1 flatly as the chip that almost killed the company — not as drama, but as accounting.

Doubling down, then waking up

Here is the part that makes the story worse before it makes it better: NVIDIA didn't immediately abandon the bet. There was a follow-up, the NV2, being developed as the graphics engine for Sega's next console — the machine that would become the Dreamcast. It, too, was built around the quadratic approach. NVIDIA was preparing to make the same wager twice, with even higher stakes, tying not one but two generations of product to a geometry the rest of the world had already voted against.

Somewhere in that stretch, the founders did the hardest thing an engineering team can do: they admitted their beautiful idea was wrong. Triangles had won. Not because they were more elegant — they weren't — but because the entire software ecosystem, led by Microsoft, had chosen them. Fighting that current meant drowning in it. The NV2 was abandoned. The quadratic religion was set down. Whatever came next would draw triangles like everybody else, and drop the expensive extras, and align with DirectX instead of arguing with it.

That clarity would eventually produce the RIVA 128. But clarity doesn't pay salaries, and by the time NVIDIA changed direction, the NV1 disaster had drained the tank. The Sega gamble had cost the company its first product and most of its runway. The relationship with Sega, though — the very partnership that had led NVIDIA down the quadratic path — was about to matter more than anyone expected. Sega's leaders had watched this scrappy Silicon Valley outfit make a catastrophic bet with real conviction, and they were still, improbably, in the room.

Which set up the most desperate stretch in NVIDIA's history: a company with a discredited architecture, a discontinued flagship, and a shrinking pile of cash, trying to build one more chip before the money ran out.

Next in the series — Part 3: "Thirty Days from Death." With the bank account nearly empty, NVIDIA bets everything on a single triangle-native chip — the RIVA 128 — and a stunning act of faith from Sega buys it just enough time.


🔗 Explore more from Syncster

Comments

Popular posts from this blog

Cursor AI Review: Is the AI Code Editor Worth It?

I've been using Cursor as my main code editor for a while now, and enough people have asked whether it's worth switching to that a proper review felt overdue. Short version: for me, yes — but with caveats. What is Cursor? Cursor is an AI-first code editor built as a fork of VS Code. That means every extension, theme, and keybinding you already use in VS Code works here, but with AI woven directly into the editing experience instead of bolted on as a plugin. It's made by Anysphere and can run models from OpenAI and Anthropic under the hood. What I like Tab completion is uncanny. Cursor predicts your next edit — not just the rest of the line, but the next change across the file. Once you get used to hitting Tab, going back to a plain editor feels slow. The Composer / Agent mode. You describe a change in plain language and it edits multiple files at once, showing you a diff to accept or reject. For refactors and boilerplate, this saves real time. It unde...

How I used Google Sheets and Apps Script

Google Sheet is one of the most powerful spreadsheet application that exists online, rivaling with Microsoft's Excel. One of the main strengths is its strong support for collaboration with other users, much easier and popular than collaboration tools with Microsoft Office. Aside from plain spreadsheet, it also supports extensions such as macro. If you are familiar with macros on other office tools, they work almost the same. However, the most extension I use and tinker with is the Apps Scipt . Apps Script Extension One of the challenges I faced recently is how do I track or monitor reports in our department if they are submitted on time or worst, forgotten due to lack of better monitoring tools. So I thought if there can be simple applications that can be deployed or use by a more general user to allow reminding periodically what reports are approaching due dates or those that are past dues. Then I looked for a way, instead of creating a full blown app from scratc...

MacBook Pro M5 vs M5 Pro: Which One Should You Actually Buy?

Apple's latest 14-inch MacBook Pro comes in two very different flavors: the base M5 and the step-up M5 Pro . On paper they look similar — same gorgeous Liquid Retina XDR display, same design — but under the hood the gap is bigger than the names suggest. Here's a clear, no-hype breakdown, with concrete use cases so you can match the chip to your work. Quick spec comparison Spec M5 M5 Pro CPU 10-core (4 performance + 6 efficiency) Up to 18-core (6 performance + 12 efficiency) GPU 10-core Up to 20-core Neural Engine 16-core 16-core Memory bandwidth 153 GB/s 307 GB/s (roughly double) Unified memory 16 / 24 / 32 GB 24 / 48 / 64 GB Max storage Up to 4 TB SSD Up to 8 TB SSD Battery (video playback) Up to 24 hours Up to 22 hours Media engines Single encode/ProRes engine More encode/ProRes engines (higher configs) What actually changes between them More cores — the M5 Pro nearly doubles CPU cores and adds GPU cores, so sustained, multi-threaded work finishe...