SambaNova
@SambaNovaAI
SambaNova is the fastest AI inference platform — running AI securely, profitably, and at scale, with ROI received at a remarkable speed that matters⚡
165 Following 13K Followers
NeurIPS week in Sydney 🇦🇺
Join SambaNova, and @intel for SambaHouse AI Night! Wed Dec 9th, 12-Micron, Barangaroo.
Lightning talks, live demos, HDI inference on sovereign AU infra, drinks on the harbor.
@v_mohan_
Show more
The slow part is writing the answer, not reading the prompt.
That's why agentic inference needs a different division of labor. GPUs for prefill, SN50 for decode.
Great breakdown of Sven's talk at the @tetrateio Inference Routing Summit!
Show more
I thought long context was what made agents slow. At @tetrateio's Inference Routing Summit, Sven Mawson of @SambaNovaAI said agent sessions hit cache 90-95% of the time, so 80-95% of the time goes to decode.
The slow part is writing the answer, not reading the prompt.
Show more
Loved having @abhiingle on the Day 1 fireside! Thanks for a great kickoff, @PlugandPlayTC 🦾
Day 1 of Silicon Valley Summit is in the books: fireside chats with former @ycombinator president @gralston and @SambaNovaAI's Abhi Ingle, plus our Linz AI Center of Excellence announcement.
Looking forward to day 2!
#PNP#
Show more
What does it actually take to build a next-generation AI inference chip?
@aton2006 sat down with @dw_dcs to go inside the SN50: the architecture decisions, the software optimization behind it, and why manufacturing scale is now part of the design problem.
Watch the full discussion here:
Show more
ICYMI: @Supermicro is building and selling SambaNova-powered inference systems.
GPUs handle prefill. SN50 RDUs handle decode. One system, built for agents, from a partner data centers already trust.
First batch: 10,000 SN50 chips.
🔗 Full blog here:
Show more
Excited to announce our collaboration with @Supermicro, the perfect partner to build and support inference racks powered by SambaNova SN50.
Our racks join Supermicro's Data Center Building Block Solutions portfolio. They fit the air-cooled data centers customers already run, next to the GPUs they already own. This first batch of SN50s is only phase one. Pumped for what comes next!
Show more
SambaNova turns 9 🥂
The team showed up and showed out at our San Jose HQ (and virtually) to celebrate this incredible milestone with us. We can't wait to keep going!
Excuse us while we fuel ourselves with some cupcakes. @RodrigoLiang @KunleOlukotun
Show more
Last week, @SambaNovaAI’s Vansanth Mohan (@v_mohan_) joined @EvanKirstel to discuss the economics of #AgenticAI#. Request count isn’t what breaks agent budgets. Output tokens per task, multiplied by generation speed, is. Because agents make serial calls, latency and cost compound together, which is why SN50 is built around decode.
Watch the full conversation below:
Show more
Always love a shoutout from our friends at @NYSEWired! We're incredibly lucky and excited to have Dan on board 🦾
Congratulations to Dan Christman on his new role as chief corporate development officer at @SambaNovaAI!
A great milestone. Well done, Dan! 👏
#NYSEWired# #theCUBE#
The request count isn’t what breaks agent budgets. It’s output tokens per task, times how fast you can generate them. Serial calls mean latency and cost compound together, which is why SN50 is built around decode. @v_mohan_ walks through the math with @EvanKirstel in their conversation. Check it out! ⬇️
Show more
AI agents turn one request into a chain of inference calls, and that's what blows up the budget. Vasanth Mohan of @SambaNovaAI broke down tokens per second per chip, model tiering, and how their RDU and disaggregated inference keep agent costs in check.
Show more
Looking for your next opportunity in AI infrastructure? 🚀
Meet SambaNova at the #SC26# Career Fair, including our EVP of Software, Rich Heaton.
All skill levels welcome! Details here:
Show more
We're partnering with @Emergys, pairing SambaNova's high-performance inference chips with enterprise deployment expertise.
"We've deployed SambaNova for our clients and seen what it does for cost and speed," said Krishna Prabhu, CEO of Emergys. "This partnership puts our services team directly behind SambaNova's customers and gives our own clients a faster path to production."
🔗Full press release:
Show more
Agents turn one request into dozens of inference calls, and that shows up fast on the AI bill. Vasanth Mohan from @SambaNovaAI explained speed per request versus throughput per chip, splitting prefill from decode, and why you should test on your own workloads.
Show more
Congrats to @General_Compute on their $100M Series A!
Proud to be part of this phenomenal team's journey 🦾
🚨 Big news today: We’ve raised $100M for our Series A! 🎉 🚀
We believe demand for fast inference is many times larger than demand for slow inference, and it’s about to go exponential.
Thanks to our co-leads @BlackbirdVC and @Abstractvc. Two visionary partners join our journey 🎉 🚀
Show more
Happening now! It's a packed house at our HQ in San Jose for AI Builders. What an incredible lineup of speakers. Thank you to everyone who attended. 🙌
@daytonaio @CipcicMarijan @v_mohan_ @youdotcom @ViniGambhir3 @muhashmii
Show more
So many great speakers tonight 😎 @daytonaio
Packed @daytonaio AI Builders in @SambaNovaAI HQ in San Jose! 🚀 Great energy tonight!
Big news: SambaNova Co-Founder and CEO @RodrigoLiang has joined the @WSJ Leadership Institute CEO Council.
He brings decades of experience building processors, systems and full-stack infrastructure for the next phase of AI.
Learn more ⬇️
Show more
As the economic center of AI shifts from training to inference, the industry's focus is moving toward sustainable deployment and achieving real financial payback for customers.
Watch @RodrigoLiang's full conversation with @furrier on @NYSEWired:
@GemmaAllenSays | @bjbaumann2014
Show more
Speed matters. And so does getting your money back.
SN50 is built for premium inference, pairing RDU decode with existing GPUs for prefill. ROI in 6 months at a 70% billable utilization planning assumption.
Read more in our blog ⬇️
Show more
