Eliyan raised $145 million at a $1 billion valuation to address AI chip data bottlenecks by improving data delivery to GPUs. The money is aimed at a very specific choke point: chips can be powerful on paper and underfed in practice. That’s the whole game here. Data has to move fast enough, or the expensive hardware sits there waiting like a state office with a broken printer.
The Bottleneck Economy
The company’s funding round was part of a broader push to ease the flow of data into AI accelerators, where bottlenecks can limit how well the chips are used. In other words, the industry keeps building bigger machines and then pays to fix the plumbing. Private capital loves that sort of problem. It gets to call the blockage innovation.
The U.S. Department of Commerce plans to award GlobalFoundries $300 million to develop faster AI chip links, including work on co-packaged optics aimed at boosting data transfer and energy efficiency. The government money is going into interconnects, the invisible hardware that lets chips talk to each other without wasting time and power. GlobalFoundries’ effort centers on interconnects that can move data faster between chips and improve how efficiently those systems use power. The language is technical. The logic is blunt. Faster links, fewer delays, more throughput, less waste.
State Money, Private Gain
The Commerce Department’s planned award shows the public sector underwriting a piece of the AI supply chain while private firms chase the rest. One side writes checks. The other side raises rounds. Both are trying to keep the machine fed. Neither is talking about who gets left out when the machine gets faster, only how to make it run with fewer hiccups.
Nvidia partner ChipAgents raised $60 million to accelerate chip design using AI agents that can make decisions and execute tasks with minimal human oversight. That pitch says a lot in a few words. The company wants software to take over more of the design workflow itself, speeding development and reducing the amount of human intervention needed. Human oversight becomes a bottleneck too. First the data. Then the links. Then the people.
Automation All the Way Down
Together, the three developments show private capital and government funding moving on separate fronts in the AI chip supply chain: data delivery to GPUs, faster links between chips, and AI-driven automation in chip design. The money is spread across the stack, from the pipes to the planning room. One round tries to unclog the flow into GPUs. Another aims to move data faster between chips. A third wants AI agents to make decisions and execute tasks with minimal human oversight.
That’s the shape of the sector right now. More funding. More automation. More attempts to shave off friction wherever a human hand still lingers. The companies call it efficiency. The government calls it development. The result is a supply chain built to move faster, think less, and keep the whole operation humming with fewer people in the loop.
Eliyan’s $145 million round, GlobalFoundries’ planned $300 million award, and ChipAgents’ $60 million raise all point in the same direction. The industry is spending heavily to solve the problems created by its own scale. First the bottleneck. Then the fix. Then the next bottleneck, waiting quietly in line.