Piris Labs

Cerebras-speed inference but scalable

Winter 2026ActiveB2BInfrastructureHard TechB2BCloud ComputingAINetworksSan Francisco, CA, USA

Company

Piris Labs offers a full-stack inference service that eliminates the AI data movement bottleneck. By pairing proprietary photonic hardware with a vertically optimized software stack, we minimize the "memory wall" associated with expensive GPUs. This allows us to deliver the same performance as traditional clusters at a fraction of the cost. Our technology improves effective FLOP utilization and reduces latency, finally making the unit economics of trillion-parameter models sustainable. Founded by a team of MIT physicists and Meta AI experts.

Verdict

High Signal

Market Opportunity

AI inference infrastructure is a massive and rapidly growing market — AI cloud compute spending is projected at $200B+ by 2030. The ICP is enterprise AI labs and hyperscalers running large-scale LLM inference who are bottlenecked by GPU memory bandwidth costs. SBIR selection signals government/defense interest as well, adding a second monetization vector.

High Signal

Founder Signal

Ali Khalatpour has a deep photonics pedigree: multiple degrees from MIT (BSc EE/Math, MASc, MSc Physics, PhD EE), led THz laser development at NASA, Head of Metasurface Compute at Imagia, lead scientist at Harvard and Stanford — 10+ years of directly relevant optical hardware R&D. Keyvan Moghadam has a USC PhD in EE, 5 years at Meta as Technical Lead/EM, 2 years as Tech Lead at X, plus a prior co-founding attempt (Esko Finance, 2023-2024). The team also added Prof. Marc Baldo (MIT RLE Director) and Mohsen Moazami (ex-Groq/NVIDIA President) as advisors/team members — serious credentialing.

Medium Signal

Competition

Direct competitors in fast inference include Cerebras (their explicit benchmark), Groq (notably, ex-Groq president joined Piris), and SambaNova. On the photonic interconnects hardware side, Ayar Labs, Lightmatter, and Celestial AI are working on similar optical interconnect approaches for AI. Piris differentiates with a full-stack vertical approach (hardware + software) rather than selling components, but the photonic-for-AI space is becoming crowded and capital-intensive.

Medium Signal

Product

Website claims 5x lower latency, 10x lower power per bit, 2x lower cost per token, but no named customers, no pricing page, no demo, and no revenue metrics are cited. They have a working 'π Conversion Engine' prototype per their October 2025 blog post and an SBIR government partnership, but this is pre-commercial — early access waitlist stage only.

OverallB Tier

Piris Labs has an unusually strong technical founder in Ali Khalatpour — his MIT/NASA/Harvard/Stanford photonics background is directly on-point and rare. The advisory additions of the MIT RLE director and ex-Groq president are meaningful validators. However, this is deep hardware/photonics infrastructure with no paying customers, no visible prototype demos, and a product that requires enormous capital to commercialize before it can displace existing GPU clusters. The SBIR win is a positive traction signal but not revenue. The memory-wall problem is real and the market is enormous, but Lightmatter, Ayar Labs, and Celestial AI are well-funded incumbents in the same photonic interconnects lane. This is a high-ceiling, high-risk bet that lives or dies on whether the hardware actually works at scale — which cannot be validated from public data today.

Active Founders

Ali Khalatpour

Founder & CEO

Founder (YC W26), reimagining AI compute architectures | PhD EE (MIT → NASA → Stanford)

Keyvan Moghadam

Founder & President

I am a technical leader with a decade of experience scaling AI/ML products. At Meta and X, I specialized in leading tiger teams to launch high-stakes 0-to-1 initiatives. Now, I am the Co-Founder of Piris Labs. We are building high-speed photonic interconnects to solve the energy and latency constraints in AI data centers. My current focus is delivering our E/O engine prototype to secure our seed round.

Piris Labs

TierB Tier

BatchWinter 2026

Team Size2

StatusActive

LocationSan Francisco, CA, USA