Liberty Global joined a funding round that values Positron at $5 billion. The company says its design avoids scarce high-bandwidth memory, but its next chip has not entered production.
New to this? Read it in simple words
- Liberty Global said on September 11 that it invested in Positron AI. It joined an $875 million funding round.
- The round values Positron at $5 billion. Positron builds systems for inference.
- Positron says its design uses common memory and needs less hard-to-find high-bandwidth memory. A design with easier-to-find memory could reduce cost.
- Its next chip is planned for production in the second half of 2027. The investment does not prove the chip will meet its goals.
- Funding round
- When investors give a young company money in return for part of it.
- Inference
- Using a trained AI model to answer questions or do tasks.
- High-bandwidth memory
- A very fast type of computer memory, often used in AI chips.
The new disclosure adds a strategic investor
Liberty Global announced its investment on September 11. It joined an $875 million funding round that values Positron at $5 billion. The round itself was announced one day earlier, so the new event is Liberty Global’s participation.
Positron builds systems for inference. Inference is the work a trained AI model does when it answers a request. This work often moves large amounts of model data from memory to a processor, which can limit speed and raise energy use.
The company calls its design memory-first. It uses LPDDR5X, a type of memory also found in many mobile and computer products. Positron says this reduces its need for scarce high-bandwidth memory and special chip packaging.
Liberty Global joined the 875-million-dollar round. Positron’s next Asimov chip is not yet in production.
The main product still has to move from plan to factory
Liberty Global says Positron’s Atlas systems already have customers, including Oracle. The new money will support a larger system called Titan and a future chip named Asimov. The company plans the first completed chip design in late 2026.
Production is planned for the second half of 2027. This means the round funds work that still carries design, factory, supply, and software risk. A successful first chip does not guarantee large production or reliable customer systems.
Positron also claims more than 90 percent use of available memory speed. That number comes from an investor announcement, not a full independent test. Buyers need results for their own models, request sizes, power limits, and software tools.
Memory competition could widen the chip market
AI chip competition often focuses on processor speed. Many real tasks are limited by how quickly memory can feed the processor. A design that uses cheaper or easier-to-find memory could reduce cost even without winning every speed test.
The trade-off may appear in power, space, delay, or the number of users served at once. Clear tests should compare full systems, not one selected number. They should include cooling, software work, and the cost of keeping machines busy.
Liberty Global’s investment gives Positron money and a possible path to large technology customers. It does not prove the next chip will meet its goals. The useful evidence will arrive through working silicon, independent tests, customer prices, and stable production in 2027.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · September 11, 2026Liberty Global Tech Ventures invests in Positron AI Liberty Global
- 2Research · September 10, 2026AI chip startup Positron’s valuation rises in latest funding round Reuters via MarketScreener
We used Liberty Global for its participation and the company’s product claims. We used Reuters for the round structure, earlier value, and market context. We separate the September 11 investor disclosure from the September 10 funding announcement.