Qualcomm nears $4 billion deal to acquire AI startup Modular

Qualcomm is closing in on a deal to acquire Modular Inc., an artificial intelligence infrastructure startup, for approximately $4 billion, according to people familiar with the matter first reported by Bloomberg. The transaction would be the chipmaker's largest AI software acquisition to date and is expected to be announced within weeks — with Qualcomm's Investor Day in New York on June 24 providing a likely backdrop for the disclosure.

What Modular Does

Founded by former Google engineers, Modular built two flagship products: Mojo, a Python-superset programming language designed for high-performance AI workloads, and MAX, an AI execution platform that optimizes how large language models run across heterogeneous multi-chip architectures. Those tools fill a critical gap in Qualcomm's portfolio: the hardware is getting there, but the software stack to deploy LLMs efficiently on non-Nvidia silicon has lagged.

Modular's valuation has climbed sharply — the company was valued at approximately $1.6 billion just nine months ago, before AI inference optimization became a top enterprise priority. The reported $4 billion price tag reflects the premium placed on any startup that can help enterprises run AI workloads outside of Nvidia's GPU ecosystem.

Qualcomm's Bigger Ambition

Under CEO Cristiano Amon, Qualcomm has been assembling the pieces of a full-stack AI platform to challenge Nvidia in the data center. The company already acquired RISC-V chip startup Ventana Micro Systems and connectivity IP provider Alphawave Semi, and confirmed that custom ASIC data-center chip shipments have been pulled forward into 2026. The Modular acquisition would add the software layer that ties the stack together.

Separately, Qualcomm is also reportedly in talks to acquire Tenstorrent — the AI chip startup backed by Hyundai and AMD — for between $8 and $10 billion. If both deals close, Qualcomm will have deployed more than $14 billion in a single year to build an end-to-end AI compute stack, a direct challenge to Nvidia's dominance from silicon to software.

What This Means for Developers

Mojo has attracted significant developer interest since its launch as a faster, hardware-aware alternative to Python for ML workloads — early benchmarks showed Mojo running certain matrix operations up to 68,000 times faster than standard Python. If Qualcomm keeps Mojo open and accelerates its development, the acquisition could meaningfully expand the ecosystem of developers writing AI code optimized for non-Nvidia hardware.

The deal is not yet finalized and terms could still change, as first reported by Bloomberg.