Qualcomm is moving to buy Modular, a Silicon Valley chip startup, in a deal valued at nearly $4 billion. The acquisition puts a spotlight on a practical but important problem in AI computing: how developers write software that can run across different kinds of chips without rebuilding it for every hardware target.
The Deal Centers on AI Software
The companies announced the acquisition on Wednesday. Qualcomm said it expects to issue up to 19.2 million shares of common stock, which works out to just under $4 billion based on the company’s last closing share price.
The deal is expected to close in the second half of this year. It also arrives nine months after Modular raised $250 million at a $1.6 billion valuation, making the purchase a sharp step up in value for the startup.
For Qualcomm, the headline number matters. But the more strategic piece is what Modular builds: a chip software platform and a proprietary coding language designed to help AI software run on different chips without requiring developers to rewrite code for each one.
What Modular Adds to Qualcomm
Modular’s core appeal is not simply that it is another chip company. The startup sits closer to the software layer that determines how developers use computing hardware.
That layer matters because AI systems often need to run across a range of compute environments. Modular’s platform is aimed at making that work less tied to one chip or one vendor’s software stack.
Qualcomm expects Modular’s full team to join the company. That includes its two cofounders and around 150 employees.
We believe the future belongs to developer-friendly, horizontal platforms that can run across diverse compute environments and give customers real choice in how and where they deploy AI
That statement from Qualcomm president and CEO Cristiano Amon frames the deal as more than an acquisition of talent. It presents Modular as part of a broader plan to make Qualcomm more relevant wherever AI software is being deployed.
Qualcomm Is Looking Beyond Mobile
The acquisition also reflects Qualcomm’s attempt to expand beyond the mobile device market, which generates the vast majority of the company’s revenues. The company is still deeply tied to mobile, but the Modular deal points toward a wider compute strategy.
Amon recently said Qualcomm has been working on 40 different chip designs for AI gadgets. Those designs include smart glasses, jewelry, earbuds, pins, and watches.
At the same time, Qualcomm has been pushing into data centers, where more powerful chips are required. That market needs hardware, but it also needs software tools that can make hardware useful to developers and customers.
Late last year, Qualcomm acquired Ventana Micro Systems, a startup focused on server CPUs based on RISC-V, an open-standard chip architecture. Qualcomm is also working on custom ASIC designs, or application-specific integrated circuits, for data centers, with China’s ByteDance reported to be an early customer.
The Founders Bring Deep Compiler and AI Experience
Modular was founded in 2022 by Chris Lattner and Tim Davis. Both worked on Google’s TPU chips before starting the company.
Davis co-created TensorFlow Lite while at Google. That version of TensorFlow allowed machine learning models to run on devices with less computing power.
Lattner’s background gives Modular a strong connection to developer tools and programming languages. He created the open source compiler infrastructure project LLVM and Apple’s Swift programming language. He was also briefly the head of Tesla’s Autopilot software program, a role later held by Andrej Karpathy, who recently joined Anthropic.
Dave Munichiello, a managing partner at GV, formerly Google Ventures, described the partnership between the cofounders as central to the company’s strength. GV was an early investor in Modular.
What makes this team truly exceptional is the complementary partnership between Chris and Tim
Chris is an N-of-1 human, in that he’s bold, visionary, and technically uncompromising.
The Competitive Question Is Portability
Lattner and Davis set out to create a unifying software layer for cloud businesses using GPUs and CPUs. The goal was to help those businesses get more out of the chips they already use, while reducing the friction of working across different systems.
That goal placed Modular in competition with important chip software ecosystems. The startup challenged Nvidia’s CUDA, a closed software system for GPUs, and AMD’s ROCm, which is open-source but not always easy to move to other chips.
Modular’s position was complicated. It eventually built partnerships with large chipmakers, hyper-scalers like Amazon, and even Apple, while also competing with those companies and the software they developed internally.
Lattner previously told WIRED that the software problem he and Davis were trying to solve was structural and had to be addressed outside a Big Tech environment. With this acquisition, that independent effort is now set to become part of Qualcomm.
The result is a deal that is about chips, but not only chips. Qualcomm is buying a team, a platform, and a developer-facing approach to AI software at a moment when running AI across many kinds of hardware has become a central challenge for the industry.