GlobalFoundries CHIPS Act funding pairs $300 million with US equity stake
The US plans $300 million for GlobalFoundries silicon photonics work while taking a 1% stake valued at about $269 million.
By Renata Fuchs · Policy Reporter
· 3 min read
The US government has signed a letter of intent for $300 million in GlobalFoundries CHIPS Act funding, alongside a plan for the government to receive a 1% equity stake in the chipmaker valued at roughly $269 million. The deal matters for AI infrastructure because the funding is aimed at silicon photonics and advanced packaging technologies used to connect increasingly large accelerator systems in datacenters.
Commerce Secretary Howard Lutnick said the investments would expand domestic semiconductor capability, support high-paying jobs and keep the US competitive in chips. The announcement was framed by the Commerce Department as part of the Trump administration’s push to strengthen US semiconductor manufacturing, rather than as a conventional private-market financing round. The letter of intent does not disclose a broader valuation for GlobalFoundries beyond the stated value of the government’s proposed stake.
What is GlobalFoundries using the CHIPS Act funding for?
GlobalFoundries says the money will support work on next-generation silicon photonics, new optical materials and advanced packaging, including 3D hybrid bonding. The company is tying the funding to domestically manufactured near-packaged optics, or NPO, and co-packaged optics, or CPO, both of which are being positioned for AI systems that need faster connections between compute components.
Silicon photonics uses optical signaling to move data, a capability that becomes more relevant as copper interconnects run into limits in large AI clusters. GlobalFoundries has already announced its Silicon Photonics Co-Packaged Advanced Light Engine, known as SCALE, which it says is designed to support 400 Gb/s per lane connectivity.
CPO places optical engines directly with the compute logic, while NPO keeps the optics in a nearby module. Both designs are expected to see commercial demand as AI system builders push beyond server-level GPU configurations into rack-scale and row-scale systems. Over the past two years, according to the reported industry shift, Nvidia and AMD system designs have moved from eight GPUs in a box to rack systems containing six dozen accelerators in one machine.
Nvidia CEO Jensen Huang underscored the market’s importance at Computex in Taipei last month, saying that optical interconnect technologies could make Marvell Technology the next trillion-dollar company. That comment reflects the broader supplier opportunity around AI networking, where bandwidth and power constraints are becoming central system-design issues.
Why GlobalFoundries is a fit for silicon photonics
GlobalFoundries has followed a different path from chipmakers chasing the most advanced CMOS logic nodes. Since its separation from AMD, the company has focused on more specialized manufacturing processes, including silicon photonics used in AI datacenters.
That positioning makes the CHIPS Act award less about leading-edge GPU production and more about the supporting infrastructure needed to scale AI compute. The government did not disclose specific production targets, customer commitments or timelines for the silicon photonics work in the announced letter of intent.
The equity component also fits a recent pattern in US semiconductor policy. GlobalFoundries would not be the first American fab operator from which the Commerce Department has sought shares in connection with CHIPS Act support. Last summer, the department converted $5.7 billion in previously awarded but undisbursed CHIPS Act grants, plus $3.2 billion from the Secure Enclave program, into an approximately 10% stake in another US chipmaker.
Intel separately said Tuesday that the Department of Defense-backed Rapid Assured Microelectronics Prototypes, Commercial program, known as RAMP-C, has ended. The program offered incentives for industry partners to build test chips in Intel fabs using the company’s 18A process technology. Intel said early access to 18A had positioned defense industrial base customers to use Secure Enclave while meeting size, weight and power requirements.
This story draws on original reporting from The Register.
