Code Metal is part of a modern wave of startups that aim to modernize the technology industry by using artificial intelligence to generate code and translate it into programming languages. However, one of the lingering questions about AI-powered code is whether the result is good and what the consequences might be if it isn’t.
Over the past two years, companies such as Antithesis, Code Rabbit, Synthesized, Theorem and Harness have attracted multi-million-dollar backing from venture capitalists for their approaches to automating, validating, testing and securing AI-generated code. These startups sell the “picks and shovels” of the AI gold rush – technology tools that serve the larger industry. While some of the methodologies behind their technology remain unproven, investors are willing to take the risk that at least some of them will succeed.
Founded in 2023, Code Metal has focused its efforts on translating and verifying codes for the defense industry. The first customers include L3Harris, RTX (formerly known as Raytheon) and the US Air Force. The startup is also working with Japanese electronics company Toshiba and says it is in talks with a major chip company to work on porting code to chip platforms, though the company wouldn’t say which one.
The startup’s software platform translates code from high-level programming languages such as Python, Julia, Matlab and C++ into lower-level languages or code that runs on specific hardware such as Rust, VHDL, and chip-specific languages such as Nvidia’s CUDA.
Code Metal CEO Peter Morales, who previously worked at Microsoft and MIT Lincoln Laboratory, says the market is starting to see “big confidence issues” in an industry that could be powered by AI-generated code in the near future. One of these problems is porting venerable code to modern applications. Morales says that if a government agency or defense contractor needs coding work done quickly but only has access to engineers specializing in an older programming language, it slows things down.
Morales quotes last post on X from renowned AI researcher Andrej Karpathy, who observed, among other things, “growing momentum for moving C to Rust.” Karpathy concluded: “It seems likely that we will repeatedly rewrite large portions of all software ever written.”
“That’s all we do in one tweet,” Morales says.
One of Code Metal’s investors, Yan-David Erlich, general partner at B Capital, says the reality is that some of the code that controls basic communications infrastructure and even satellites “is old, sloppy, written in programming languages that people may no longer use. It needs to be modernized.”
“But as you translate,” Erlich added, “you can insert errors, which is disastrously problematic.”
The startup is afraid to share too many details about its methodology. However, one element of the business that he is not afraid to talk about is his approach to pricing.