About the name: Vertex

The name comes from computer graphics, not branding. A vertex is the point where the edges of a triangle meet, and the triangle is the basic building block every 3D surface a GPU renders is made of. It's the literal reason NVIDIA exists as a company: long before AI or data centers, NVIDIA built chips to push triangles onto a screen faster than anyone else.

Why this exists

NVIDIA's own site splits its lineup across dozens of separate product pages. No single resource, official or third-party, pulls the whole hardware and software catalog into one place at the same depth. Vertex closes that gap: one table, described the same way, so products can be compared instead of chased down one at a time. Every entry carries a plain-English explanation next to a technical one, a maturity status, a release date, and named competitors.

It also solves a second problem: NVIDIA adds new products most months, and it's easy for a team's shared understanding to drift out of date. Vertex is built to double as training material and as a common reference for customer and partner meetings, so everyone walks in speaking the same language about what a product actually is, not everyone's own half-remembered version of it.

Who it's for

IT professionals, solution architects, infrastructure engineers, and developers who work with NVIDIA hardware or software and need a working grasp of the full catalog without living inside it every day. That covers someone scoping a GPU cluster and weighing DGX against HGX against MGX, an architect deciding which networking or virtualization layer a deployment needs, a developer figuring out whether NIM, Triton, or a custom serving stack fits, and anyone prepping for a customer or partner conversation who wants the plain facts before the sales pitch.

How to use it

• Scanning a category. The catalog groups everything by category — silicon, systems, networking, edge/robotics/automotive, core AI software, verticals, cloud, infrastructure management, dev tools. Jump to the one that matches what you're working on.
• Filtering by relevance. Use the Tier filter (1, 2, 3) to cut straight to what matters for most enterprise/AI-infrastructure work versus what's reference-only. Tier 1 rows are worth knowing cold.
• Explaining it to someone else. Every entry has a Plain-English version, how you'd explain it to a sharp 10-year-old, and a Technical Deep-Dive version, for someone technical but outside that specific domain. Pick whichever fits the room — a training session or a customer meeting usually wants both.
• Checking maturity. Status and Release Date flag anything that isn't fully GA yet, so a design or pitch doesn't accidentally lean on something still sampling.

Value

• One place instead of dozens of NVIDIA product pages, described at consistent depth and in consistent format.
• Fact-checked against NVIDIA's own sources, not general knowledge, with corrections flagged inline where the record needed fixing.
• Built for training sessions and customer or partner meetings: plain-English and technical explanations side by side, competitors named, maturity called out honestly instead of glossed over.
• A living reference, kept current as NVIDIA ships more — see the Versioning tab for what changed and when.

How to read an entry

Tier

A priority ranking for enterprise AI-infrastructure work, not a ranking of NVIDIA's business importance.

• Tier 1Comes up in nearly every AI-infrastructure or enterprise conversation. Worth knowing cold.
• Tier 2Matters for specific verticals or deployment types. Good to know exists.
• Tier 3Reference-only: consumer, legacy, or niche. Unlikely to come up unless specifically asked.

Status

• GAGenerally available.
• Limited AvailabilitySampling, just-launched, or regionally rolling out. Not yet broadly shipping.
• Beta / PreviewNot GA: early access, technical preview, or open-source-in-progress.
• Pending AcquisitionAnnounced but not yet closed.
• Legacy / SunsetStill around, being phased out.

Everything else on a card

Release date uses year-month where NVIDIA has confirmed one; a "~" marks a date inferred from indirect sourcing rather than an exact NVIDIA-stated date. Plain-English is how you'd explain the product to a sharp 10-year-old. Technical deep-dive is the same product explained to someone technical but outside that specific domain. Competition names the closest real alternatives, or says so when a product has no direct rival.

Feedback

Vertex is a living reference. If something looks outdated, missing, or wrong, or you have a suggestion, send it to vertex@rspruijt.com so it can be corrected.