The US government is betting $300 million that light-based data transfer can break the AI data center bottleneck.
The US government is betting $300 million that light-based data transfer can break the AI data center bottleneck.
The US government is betting $300 million that light-based data transfer can break the AI data center bottleneck.
GlobalFoundries will receive $300 million from the US Commerce Department to develop silicon photonics technology, a light-based data transfer method that promises 5x better energy efficiency for AI data centers.
"GlobalFoundries has spent more than a decade building the technology, footprint and ecosystem to lead this transition, and we have the proven manufacturing foundation to scale it — in the United States," Chief Executive Officer Tim Breen said.
The company targets data transfer speeds of 400 gigabits per second using co-packaged optics, which integrate silicon photonics directly alongside AI processors. Much of the supply chain for these technologies is currently concentrated outside the US, with Taiwan Semiconductor Manufacturing Co. and Israel's Tower Semiconductor dominating production.
The award, part of President Donald Trump's push to direct CHIPS Act research funding toward critical semiconductor technologies, strengthens GlobalFoundries' position in a market where data center interconnect spending is projected to exceed $20 billion annually by 2030. The company's shares trade at roughly 22x forward earnings, a discount to TSMC's 28x, reflecting its narrower focus on specialty nodes rather than leading-edge logic.
Silicon photonics replaces electrical signals with light to move data between chips, addressing a growing bottleneck in AI systems where compute performance has outpaced data transfer speeds. The technology is particularly critical for large-scale AI training clusters that link thousands of accelerators, where interconnect bandwidth and power consumption directly affect training time and cost.
GlobalFoundries will conduct the research at its facilities in Malta, New York, and Burlington, Vermont. The company generated $542 million in cash from operations in the first quarter of 2026 and spent $309 million on capital expenditures, or roughly 19% of revenue. It expects net capital expenditure to increase in the second quarter, driven by tool delivery timing and government grants.
The award adds to a series of CHIPS Act disbursements under the Trump administration, which previously committed $150 million for semiconductor manufacturing equipment and $2 billion for quantum computing. The Commerce Department has now allocated roughly $35 billion of the $52.7 billion CHIPS Act fund, with the remaining balance directed toward future R&D programs.
TSMC, the dominant player in advanced packaging, has invested heavily in its own silicon photonics and co-packaged optics capabilities. The Taiwanese foundry allocated 10% to 20% of its $60 billion to $64 billion 2026 capital budget to advanced packaging, testing and mask-making. GlobalFoundries, by contrast, focuses on specialty technologies where it holds stronger competitive positioning, including silicon photonics, FDX (a low-power platform) and high-performance silicon-germanium.
The US government's investment aims to reduce reliance on Asian supply chains for optical packaging, a segment where TSMC and Tower Semiconductor currently hold the manufacturing lead. For investors, the question is whether GlobalFoundries can convert this R&D funding into commercial revenue before TSMC's packaging scale advantage widens further. GlobalFoundries reaffirmed its full-year 2026 non-IFRS net capital expenditure outlook of 15% to 20% of revenue, a range that keeps spending in check even as it pursues new growth corridors.
This article is for informational purposes only and does not constitute investment advice.