If vLLM already solved LLM serving, why did SGLang appear?
Meet Neysa at Cypher
Neysa is the AI Cloud Partner at Cypher 2026, the 10th edition of India’s AI conference by Analytics India Magazine, held 7 to 9 October 2026 at KTPO, Whitefield, Bengaluru. Find us at Booth 3 on the expo floor all three days, showcasing the all-new Neysa Velocis experience, and catch Ritesh Dhoot’s talk on the real cost of production AI on 7 October at 4:45 PM IST in Hall 1.
At Cypher, we’ll explore what it really takes to run AI in production, from inference economics and infrastructure utilisation to model selection, workload architecture and operational control.
Connect with the Neysa team to see how the right infrastructure decisions can help move AI from experimentation to production.
What We’re Bringing
Experience the full Velocis stack live, from bare-metal silicon to served inference endpoints.
Spin up pre-configured Jupyter or VS Code environments in seconds. Provision runtimes on-demand: bare-metal, VMs, managed K8s, or dedicated inference endpoints.
See how infrastructure decisions such as caching context, matching models to workloads, separating batch and real-time traffic, and choosing dedicated capacity can change the economics of production AI.
Neysa on Stage
Ritesh Dhoot, EVP Engineering at Neysa, addresses a critical question for engineering teams scaling production AI.
Stop counting tokens. Start measuring outcomes.
Where your inference runs decides what it costs.
7 October 2026 | 4:45 PM to 5:05 PM IST | Hall 1 | 20 minutes including Q&A
Speaker: Ritesh Dhoot
The published price per token is easy to compare. The real AI bill is not.
As AI moves from experimentation to production, the way we think about its economics needs to evolve.
Ritesh will explore what really drives the cost of running AI at scale, and why measuring success purely through tokens can miss the bigger picture.
The conversation will look beyond the price sheet to cost per outcome, and the role that inference, infrastructure and workload architecture play in shaping the economics of production AI.
If you’re building, scaling or deploying AI, this is a conversation you won’t want to miss.
Join Ritesh at Cypher to rethink how you measure the true cost of AI.
Coming to Cypher? Let’s Meet.
Whether you’re building AI applications, scaling inference workloads or evaluating where your enterprise AI should run, we’d like to understand what you’re working on.
Book a meeting with Neysa to discuss your AI workload, infrastructure requirements, cost per outcome and the path from experimentation to production.
FAQs
What is Neysa’s role at Cypher?
Neysa is the AI Cloud Partner at Cypher, engaging with AI builders, technology leaders and enterprises around the infrastructure required to run AI workloads at scale.
What will Ritesh Dhoot’s session cover?
Ritesh will explore the real cost of production AI – moving beyond token pricing to repeated context, capacity utilisation, data movement, model choice and infrastructure architecture. He will also show why cost per outcome is a more useful measure for production workloads.
Why isn’t the published price per million tokens enough?
Token pricing captures only one component of the overall cost. AI workloads can also involve repeated conversation history, multiple model calls, infrastructure capacity, data movement and other operational costs.
Is running open-weight models always cheaper?
No. Running open-weight models can provide greater control over infrastructure and workload economics, but organisations also take on the cost of hardware, utilisation and the people required to operate it. The economics depend on the workload and volume.
How can I calculate what my AI workload actually costs?
Start with real traffic from a single workload over a month. Measure requests, actual tokens processed, infrastructure utilisation and data leaving your environment. Then factor in retries, human fixes, idle capacity and AI applications purchased directly by business teams to build a more complete picture of cost per successful outcome.
Where is Neysa’s booth at Cypher 2026?
Neysa is at Booth 3 on the Cypher 2026 expo floor at KTPO, Whitefield, Bengaluru, all three days from 7 to 9 October 2026. Drop by for live demos of Neysa Velocis, from bare-metal GPUs to served inference endpoints.
How do I book a meeting with Neysa at Cypher?
Use the meeting form on this page to share your AI workload and preferred day. The Neysa team will confirm a slot at Booth 3 to discuss your infrastructure requirements, cost per outcome and the path from experimentation to production.