内容来源：https://www.wired.com/story/vibe-coding-startup-code-metal-raises-series-b-fundraising/

内容总结：

波士顿AI编程创企Code Metal完成1.25亿美元B轮融资，瞄准国防工业代码现代化需求

总部位于波士顿的AI编程初创公司Code Metal近日宣布完成1.25亿美元B轮融资，由Salesforce Ventures领投，Accel、B Capital、J2 Ventures等新老投资者跟投。此次融资距其获得由Accel领投的3600万美元A轮融资仅过去数月，公司投后估值已达12.5亿美元。

Code Metal成立于2023年，专注于利用人工智能实现代码在不同编程语言间的自动翻译与验证，其核心业务聚焦于国防工业领域。公司已与L3Harris、RTX（原雷神公司）及美国空军等客户展开合作，并与东芝等电子企业达成协作，同时正与一家大型芯片公司就跨平台代码移植进行洽谈。

公司平台可将Python、Julia、Matlab、C++等高级语言代码，转换为Rust、VHDL或英伟达CUDA等硬件专用低级语言。CEO彼得·莫拉莱斯指出，随着AI生成代码日益普及，如何将遗留系统代码高效移植至现代环境已成为行业关键挑战。B Capital普通合伙人严大卫·埃利希也强调，许多关键基础设施的代码"陈旧且基于过时语言，亟需现代化改造"。

针对代码翻译可能引入错误的风险，Code Metal称其专利技术能在每个转换步骤中生成测试工具链，实时验证代码可靠性。莫拉莱斯表示，对于当前已建立的转换流程，"系统不会产生错误，若无法完成翻译会直接提示无解"。

在商业模式上，Code Metal摒弃了传统按席位收费的模式，转而根据内核开发时长、代码行数转换量或节省的开发时间等价值指标进行定制化定价。为拓展企业市场，公司近期任命前Tableau CEO瑞安·艾泰为总裁兼首席运营官，并聘请前美国国家安全委员会中国事务主任劳拉·沈担任增长执行副总裁。

随着Antithesis、Code Rabbit等一批专注于AI代码验证测试的初创企业获得资本青睐，行业正形成围绕AI编程工具的"淘金热卖铲人"生态。Code Metal宣称已实现盈利，成为兼具高估值与正向现金流的"双料独角兽"。在AI重构软件开发格局的浪潮中，这类企业能否将技术优势转化为长期竞争力，仍有待市场检验。

中文翻译：

波士顿初创企业Code Metal近日完成1.25亿美元B轮融资，新老投资者共同参与。这家公司专注于利用人工智能编写代码并将其转换为其他编程语言，距离其获得Accel领投的3600万美元A轮融资仅过去数月。

Code Metal代表着以AI革新科技行业的新浪潮，其核心是通过人工智能生成代码并实现跨编程语言转换。但围绕AI辅助代码的质疑始终存在：生成代码的质量究竟如何？如果质量不佳会引发什么后果？

过去两年间，Antithesis、Code Rabbit、Synthesized、Theorem、Harness等公司凭借其AI代码自动化、验证、测试与安全保障方案，已从风险投资机构获得数千万美元融资。这些初创企业正在出售AI淘金热中的"镐与铲"——服务于更广阔行业的技术工具。尽管部分技术方法尚未经过验证，投资者仍愿意押注其中至少有些能获得成功。

成立于2023年的Code Metal将业务重点聚焦于国防领域的代码翻译与验证，其早期客户包括L3Harris、RTX（原雷神公司）和美国空军。该公司还与日本东芝电子开展合作，并透露正与某大型芯片公司洽谈跨芯片平台代码移植项目，但未透露具体企业名称。

该公司的软件平台能将Python、Julia、Matlab、C++等高级编程语言代码，转换为Rust、VHDL等低级语言或特定硬件运行代码，包括英伟达CUDA等芯片专用语言。

曾在微软和MIT林肯实验室任职的Code Metal首席执行官彼得·莫拉莱斯指出，市场正开始认识到这个行业面临的"支柱性难题"——在不远的未来，这些难题可能需要依靠AI生成代码来解决。其中关键挑战在于将旧代码移植到新应用。莫拉莱斯举例说，若政府机构或国防承包商急需完成编码工作，却只能找到精通过时编程语言的工程师，整个进程就会严重滞后。

莫拉莱斯引用知名AI研究员安德烈·卡帕西近期在X平台的发文，其中特别提到"将C语言移植到Rust的趋势正加速形成"。卡帕西总结道："我们很可能需要将历史上所有软件的大部分代码进行多轮重写。"莫拉莱斯对此回应："这条推文完美概括了我们的全部工作。"

B Capital普通合伙人、Code Metal投资人严大卫·埃利希指出，现实情况是控制关键通信基础设施乃至卫星系统的部分代码"已经陈旧过时，采用现今可能无人使用的编程语言编写，亟需现代化改造"。但他补充道："在翻译过程中可能会引入错误——这将导致灾难性问题。"

Code Metal宣称其专有技术正是为此而生。莫拉莱斯表示，在翻译过程的每个环节，公司软件都会生成系列测试工具——包含数据和工具的虚拟容器——用以评估代码性能并向客户实时展示运行状态。被问及代码翻译错误率时，莫拉莱斯称主要取决于代码转换难度，但对于公司当前运行的流程管道，"根本不会产生错误。如果无法完成翻译，软件只会提示'无解方案'。"

这家初创企业对技术细节披露持谨慎态度，但在定价策略上却毫不讳言。当"按席位收费"模式逐渐过时之际，Code Metal与数千家试图销售企业软件的AI初创企业正面临转型。甲骨文、Salesforce等企业软件巨头长期根据客户员工数量定价，而在生成式AI时代，数据处理单元成本高昂，"节省时间"成为关键衡量指标，企业正试图就新的价值度量标准达成共识。

莫拉莱斯透露公司采用定制化定价谈判，通常基于三大要素之一：内核开发耗时、代码行翻译量（基于编写代码时间）、或节省的开发时间。换言之，公司不仅按数据处理单元成本收费，更注重使用其软件实际节省的时间——这种衡量方式存在一定弹性空间。（莫拉莱斯承认价值确定过程可能"变得模糊"，但表示目前运行良好，且"所有试点项目最终都进入了下一阶段"）

作为销售战略部署，公司聘请前Salesforce旗下Tableau应用首席执行官瑞恩·艾泰担任总裁兼首席运营官，去年还招募前美国国家安全委员会中国事务主任劳拉·沈担任增长执行副总裁。

本轮由Salesforce Ventures领投，Accel、B Capital、J2 Ventures等机构参与的1.25亿美元融资，使Code Metal在私募市场的估值达到12.5亿美元。公司宣称已实现盈利，成为多重意义上的独角兽——估值超十亿美元且产生正向现金流。在AI新格局中，这种优势能否转化为长期成功，仍是所有志在突围的初创企业面临的共同课题。

英文来源：

Code Metal, a Boston-based startup that uses AI to write code and translate it into other programming languages, just closed a $125 million Series B funding round from new and existing investors. The news comes just a few months after the startup raised $36 million in series A financing led by Accel.
Code Metal is part of a new wave of startups aiming to modernize the tech industry by using AI to generate code and translate it across programming languages. One of the questions that persists about AI-assisted code, though, is whether the output is any good—and what the consequences might be if it’s not.
Over the past two years companies like Antithesis, Code Rabbit, Synthesized, Theorem, and Harness have all secured millions in backing from venture capitalists for their approaches to automating, validating, testing, and securing AI-generated code. These startups are selling the “picks and shovels” of the AI gold rush—tech tools that serve a larger industry. While some of the methodologies behind their technology remain unproven, investors are willing to gamble that at least a few will pan out.
Code Metal, which was founded in 2023, has focused its efforts on code translation and code verification for the defense industry. It boasts L3Harris, RTX (formerly known as Raytheon), and the US Air Force as early customers. The startup is also working with Japanese electronics company Toshiba and says it’s in talks with a large chip company to work on code portability across chip platforms, though the company declined to say which one.
The startup’s software platform translates code from high-level programming languages like Python, Julia, Matlab, and C++ to lower-level languages or code that runs on specific hardware, like Rust, VHDL, and chip-specific languages like Nvidia’s CUDA.
Code Metal CEO Peter Morales, who previously worked at Microsoft and the MIT Lincoln Laboratory, says the market is starting to recognize “the big tentpole problems” in an industry that could, in the not-so-distant future, be propped up by AI-generated code. One of those problems is porting old code into new applications. If a government agency or defense contractor needs coding work done quickly, Morales says, but only has access to engineers who have specialized in a legacy programming language, that slows everyone down.
Morales cites a recent post on X from well-known AI researcher Andrej Karpathy, who observed the “rising momentum behind porting C to Rust,” among other things. Karpathy concluded: “It feels likely that we’ll end up rewriting large fractions of all software ever written many times over.”
“That is all of what we do in one tweet,” Morales says.
One of Code Metal’s investors, Yan-David Erlich, a general partner at B Capital, says the reality is that some of the code that controls essential communications infrastructure, and even satellites, “is old, it’s crufty, it’s written in programming languages that people might not use anymore. It needs to be modernized.”
“But in the course of translation,” Erlich added, “you might be inserting bugs—which is catastrophically problematic.”
That’s where Code Metal says its proprietary tech comes in. Morales says that at each step of translation, Code Metal’s software generates a series of test harnesses—a virtual container of data and tools—that evaluate the code and show customers along the way that it’s working. When asked about Code Metal’s error rate for translation, Morales says it depends largely on how difficult the code conversion is, but that for the pipelines Code Metal currently runs, “there’s no way to generate an error. The software will just say, ‘There’s no solution for this’ if we can’t complete the translation.”
The startup is skittish about sharing too many details about its methodology. One element of the business it’s not shying away from talking about, however, is its approach to pricing.
Code Metal, along with thousands of other AI startups trying to sell enterprise software, is coming of age at a time when the “per seat” sales model is growing stale. Enterprise software giants like Oracle and Salesforce have long sold their services to customers based on how many employees that customer has. In the age of generative AI, when tokens—the units of data being processed—are expensive and “time saved” has become a critical measurement, companies are instead trying to agree on new metrics of value.
Morales says that his company negotiates pricing with every customer individually and that the cost is typically based on one of three factors: the amount of time it takes to develop a kernel, lines of code translated (which is based on time to write code), or development time saved. In other words, the company isn’t just charging for the raw cost of tokens but the actual time saved by using the company’s software, which can get a little squishy. (Morales acknowledges that the process of determining value can “get murky” but says it seems to be working so far and that “every pilot [the company has] deployed ends up going to the next phase.”)
As part of its sales strategy, the startup has hired executive Ryan Aytay as its president and chief operating officer. Aytay was previously at Salesforce, where he was chief executive of the Salesforce-owned app Tableau. Laura Shen, previously the director for China at the US National Security Council, was brought in last year as the company’s executive vice president of growth.
With this latest $125 million funding round, which was led by Salesforce Ventures and includes Accel, B Capital, J2 Ventures, and others, Code Metal is now valued on the private market at $1.25 billion. The company claims it’s profitable, making it a unicorn in more ways than one—valued at over $1 billion and already generating positive cash flow. Whether that translates to long-term success remains a question for any startup looking to stake its claim in the new AI landscape.