- Neurophos develops Tulkas T100 optical processor capable of 470 petaFLOPS AI calculations
- Optical transistors are 10,000 times smaller than current conventional silicon photonics
- Dual grid design integrates 768GB of HBM for memory-intensive workloads
Austin-based startup Neurophos has revealed that it is hard at work developing an optical processing unit called the Tulkas T100 that promises major advances in computing.
Funded by Bill Gates’ Gates Frontier Fund, the company claims the chip can deliver 470 petaFLOPS of FP4 and INT4 compute while consuming between 1 and 2 kW under load.
Its optical tensor core measures approximately 1000 x 1000, which is approximately 15 times larger than the standard 256 x 256 arrays used in current AI GPUs.
Optical transistors and extreme speeds.
Neurophos optical transistors aim to push the limits of traditional semiconductors by extending Moore’s Law through increased compute density without increasing power consumption or chip size.
Despite its scale, the startup says it requires just one core per chip, backed by ample RAM and vector processing units to maintain performance.
Its optical transistors are approximately 10,000 times smaller than current silicon photonic components, allowing a high-density die to fit onto a single reticle-sized die.
“The equivalent of the optical transistor you get today in Silicon Photonics factories is huge. It’s like 2mm long,” said Neurophos CEO Patrick Bowen.
“It’s not possible to put enough of them on a chip to get computing density that remotely competes with today’s digital CMOS.”
The Tulkas T100 runs at 56 GHz, far exceeding previous CPU and GPU clock speeds.
SRAM powers the tensor core to maintain efficiency and SSD storage can help move large data sets during testing and simulation.
The chip uses a dual-lattice design with 768GB of HBM to support memory-intensive AI workloads.
Neurophos says the first-generation Tulkas T100 will focus on the preloading stage of AI inference by handling input token processing for large language models.
Bowen envisions combining Tulkas chip racks with existing AI GPU racks to accelerate computing.
However, the company doesn’t expect full production until mid-2028, with initial shipments in the thousands.
Engineers are currently testing a proof-of-concept chip to validate the claimed computing density and power consumption.
Competitors such as Nvidia and AMD are also investing heavily in silicon photonics, indicating growing competition in this field.
AI tools and memory bandwidth limitations remain central considerations as optical accelerators look to complement conventional GPUs.
While the Tulkas T100 shows potential to advance AI computing, its practical impact remains uncertain until the company achieves reliable production.
The optical approach remains experimental and faces challenges related to SRAM requirements, vector processing, and CMOS manufacturing integration.
Optical transistors could speed up matrix multiplication and reduce power per operation, but effectiveness depends on memory, SSD storage, and AI integration.
Neurophos claims its chips are compatible with standard semiconductor fabs, but mass production depends on solving these engineering challenges.
Through The Registry
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds. Be sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp also.
