Announcing VectorWare – VectorWare

Our thesis

Technology shifts happen gradually, then suddenly. We are in the suddenly part. New
technologies like LLMs, generative AI, self-driving cars, drones, AR/VR, and robots are
reshaping the world. But they are not the technology shift. They are the new
applications enabled by it. The real shift is from CPU to GPU.

The importance of CPUs and GPUs has inverted. To compete, CPUs are adding GPU
features while GPUs are adding CPU features. CPUs and GPUs are converging.

Software has not kept pace. CPU software is advanced, standardized, and familiar. GPU
software is primitive, bespoke, and weird. Most programmers still focus on the CPU.

We believe we are at the start of a new software industry. We intend to lead it.

GPU-native software

There are two broad classes of applications:

1. GPU applications such as AI, computer vision, machine learning, scientific
   simulations, and graphics. These require GPUs and are driving most demand,
   investment, and improvements in compute hardware today.
2. CPU applications, which includes nearly everything else.

If you look at existing GPU applications, their software implementations aren’t truly
GPU-native. Instead, they are architected
as traditional CPU software with a GPU add-on. For example,
pytorch uses the CPU by default and GPU acceleration is
opt-in. Even after opting in, the CPU is in control and orchestrates work on the GPU.
Furthermore, if you look at the software kernels that run on the GPU they are simplistic
with low cyclomatic complexity. This
is not unique to pytorch. Most software is CPU-only, a small subset is GPU-aware, an
even smaller subset is GPU-only, and no software is GPU-native.

We are building software that is GPU-native. We intend to put the GPU in control. This
does not happen today due to the difficulty of
programming GPUs, the immaturity of GPU software and abstractions, and the relatively
few developers targeting GPUs.

With the advent of GPU databases, we are just starting to see CPU-based applications
migrate to GPUs. As CPUs and GPUs converge, we believe that all software will begin to
leverage GPUs to varying degrees. This is a huge opportunity.

At VectorWare we are excited to focus on both improving GPU applications and migrating
CPU applications to the GPU. We are building supporting tools and a new low-level
software stack to make GPU-native software a reality.

Think of us like:

Spreadsheets

Killer app making the new hardware ubiquitous

=

AI

Killer app making the new hardware ubiquitous

Creates platforms, apps, and developer tools for the ubiquitous hardware

=

VectorWare

Creates platforms, apps, and developer tools for the ubiquitous hardware

Who we are

Our company is comprised of Rust compiler team members, open source maintainers of
rust-gpu,
rust-cuda, and
rustc_codegen_clr, as well as
graphics experts from the gaming industry. In the past we’ve worked on everything from
operating systems at Apple, browsers at Mozilla, web and mobile apps at Facebook, and
graphics technology at Embark Studios and Frozenbyte. We’ve led developer tools and
infrastructure teams and even built our own
IDE
long before similar tools became billion-dollar AI companies. You can read more about us
on our team page.

We had overwhelming interest from investors and a heavily oversubscribed seed round.
Ultimately, we chose to raise a smaller amount from people we know well and have worked
with at previous companies. We met Dan Portillo, co-founder of The General
Partnership, while working at Mozilla and are thrilled to have
him as our lead investor. Our angel investors include:

• John Lilly, an
  experienced investor, operator, and leader. We worked with him at Mozilla where he was
  the CEO.
• Patrick Kavanagh, one of the first
  angel investors in Robinhood and an early investor in hot AI startups such as
  Manus and Plaud. We worked with him at
  Robinhood where he was the head of international and crypto.
• Nick Candito, a career entrepreneur who has
  seen three early-stage ventures scale to nearly $900M in acquisition value and has
  been part of over 300 private investments ($75M allocated, ~20 unicorns, 15+ exits, 10
  funds). We met him when he was founding Progressly (later acquired by
  Box).

These folks are experienced investors as well as founders and operators who understand
the challenges of building. We’re grateful they chose to invest their time and money in
us.

We’re hiring

We are growing our early team and are hiring for a few key
roles.

GPU-native application engineering

• Goal: Ship GPU-native applications and build the missing abstractions that make
  them feel ordinary. Write “X for the GPU” where X is virtually any application.
• Ideal background: Rust expertise plus experience with GPUs (CUDA, Vulkan, ROCm,
  CANN) and/or machine learning. Alternatively, the creator or maintainer of widely used
  Rust software with an interest in learning about GPUs.
• Also welcome: GPU or ML experts who want to learn Rust.

Compiler engineering & language design

• Goal: Shape the low-level stack and language features that keep GPU-native software
  safe, performant, ergonomic, and reusable.
• Ideal background: Contributor to the Rust compiler, preferably including wasm,
  Cranelift, or LLVM. Or experience writing implementations of other languages or
  emulators in Rust.
• Also welcome: Language or tooling experts (wasm, Triton, LLVM, MLIR, Mojo, shader
  compilers) ready to learn Rust.

Userland graphics engineering

• Goal: Modify the graphics stack that GPU-native applications depend on to improve
  the safety, performance, ergonomics, and reusability of GPU-native applications. This
  includes APIs like Vulkan, plus stacks such as Mesa, DRM, Wayland, llvmpipe, MoltenVK,
  and KosmicKrisp.
• Ideal background: Rust and graphics experience with a deep understanding of GPU
  APIs and architectures or compatibility layers.
• Also welcome: Graphics engineers who want to learn Rust.

Linux kernel engineering

• Goal: Push the OS to better support GPU-native applications, improving safety,
  performance, ergonomics, and reusability from the kernel up when running in the
  datacenter.
• Ideal background: Linux kernel developers working on Rust-based graphics, storage,
  or networking drivers. Working directly on Rust for Linux would be great too.
• Also welcome: Seasoned Linux kernel engineers who want to learn Rust and GPUs.

For more information and to get in touch, please visit our jobs
page.