Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Code Metal, A A Boston-based startup that uses artificial intelligence to write code and translate it into other programming languages has just closed a $125 million Series B funding round from new and existing investors. This news comes just a few months after the startup raised $36 million in Series A funding led by Accel.
Code Metal is part of a new wave of startups aiming to modernize the tech industry by using artificial intelligence to create code and translate it across programming languages. However, one question that remains about AI-powered code is whether or not the result is good, and what the consequences might be if it is not.
Over the past two years, companies like Antithesis, Code Rabbit, Syntheised, Theorem, and Harness have received backing from millions of venture capitalists for their approaches to automating, validating, testing, and securing AI-generated code. These startups are selling the “shovels and shovels” of the AI gold rush, technical tools that serve a larger industry. While some of the methodologies underlying this technology remain unproven, investors are willing to gamble that at least a few of them will work.
Code Metal, founded in 2023, has focused its efforts on code translation and code verification for the defense industry. It boasts L3Harris, RTX (formerly Raytheon), and the US Air Force as early customers. 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 the code across chip platforms, though the company declined to specify which.
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 recognize “big problems” in the industry that could, in the not-too-distant future, be supported by AI-generated code. One of these problems is porting legacy code to new applications. If a government agency or defense contractor needs to get coding work done quickly, but only has access to engineers who specialize in a legacy programming language, he says that slows everyone down.
Morales quotes Recent post on X From renowned AI researcher Andrei Karpathy, who noted “the growing momentum behind porting C to Rust,” among other things. “We’ll probably end up rewriting significant portions of all the software that’s ever been written multiple times,” Karpathy concluded.
“That’s all we do in one tweet,” Morales says.
The reality is that some of the code that controls basic communications infrastructure, even satellites, is “old, clumsy, and written in programming languages that people may not use anymore,” says Code Metal investor Jan David Ehrlich, a general partner at B Capital. “It needs to be updated.”
“But during translation, you might introduce errors, which is a catastrophic problem,” Erlich added.
This is where Code Metal says its technology comes in. At each step of translation, Code Metal creates a series of test tools — a virtual container for data and tools — that evaluate the code and show customers along the way it works, Morales says. When asked about Code Metal’s compilation error rate, Morales said it largely depends on how difficult the code is to compile, but for the pipelines Code Metal currently runs on, “there’s no way to generate an error. The software will just say, ‘There’s no solution to this,’ if we can’t complete the compilation.”