PicoJool has raised $27.5 million to commercialize a new generation of optical interconnect technology aimed at one of AI infrastructure’s fastest-growing bottlenecks: moving data between accelerators.

The September 24 Series A was led by Socratic Partners with participation from Hudson River Trading. Combined with an earlier $12 million seed round led by Playground Global, PicoJool has now raised $39.5 million.

The Palo Alto company plans to expand teams and facilities in the United States and Taiwan while qualifying and manufacturing its 200G VCSELs, microVCSELs and optical-link products for hyperscalers and data-center operators.

From 800G to 1.6T and 3.2T

VCSEL stands for vertical-cavity surface-emitting laser. PicoJool’s portfolio includes 100G and 200G high-speed VCSEL products plus lower-power 50G NRZ and 64G NRZ/PAM4 microVCSEL configurations. The company says its 200G devices exceed 37 GHz of bandwidth.

Rather than treating 1.6T as a single architecture, PicoJool offers multiple paths. Quad 200G devices can emphasize lane speed, while a 32×50G NRZ microVCSEL configuration uses massive optical parallelism. The company is targeting optical connectivity scaling from 800G through 1.6T, 3.2T and beyond.

The embedded Semi Doped interview with PicoJool CEO Al Yuen is in English and directly examines GaAs VCSELs for AI scale-up interconnects, including the roadmap from 1.6T toward 3.2T and 12.8T. The specific YouTube watch URL is used so WordPress can render its native playable oEmbed. Voice translation is not required because the interview is already in English.

AI is pushing optics closer to compute

As accelerator counts rise, bandwidth and energy per transmitted bit increasingly constrain cluster scale. PicoJool is developing active optical cable, near-packaged-optics and co-packaged-optics approaches designed to move optical connectivity closer to the processors.

The company is working with WIN Semiconductor and other GaAs foundries to prepare its 200G VCSELs for production. PicoJool says it has begun early sampling of chip-level products while working with system companies and hyperscalers on future pluggable, NPO and CPO designs.

This complements BitcoinVersus.tech’s recent coverage of Marvell’s 3.2T AI networking roadmap and 1.6T optical engines moving toward 6.4T links.

Read PicoJool’s September 24 funding announcement. Independent coverage from SiliconANGLE and SDxCentral also reported the round and the company’s optical roadmap.

BitcoinVersus.Tech

Advertisement: Use promo code bitcoinversus at checkout to get 5% off participating Bitaxe mining products.

BitcoinVersus.Tech Editor’s Note & Disclaimer:
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. BitcoinVersus.tech provides news, technical analysis and educational information and does not provide financial or investment advice. Manufacturer performance claims should be independently verified before purchasing or deploying hardware. If you would like to support to help further secure the integrity of our research initiatives, please donate here: 3C9o19EH5HSiwEPyCTmEKzxhNCbo2X6TTb