The Pulse
Nvidia Joins Simplismart’s $14M Series B for AI Inference Software
Simplismart raised $14 million in a Series B led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC and Micromax participating. The Bengaluru startup says it will use the funding to develop inference software and expand partnerships
AI.info Team · Written with AI · fact-checked ·
Simplismart says its annualized revenue run rate grew 14-fold in six months. The Bengaluru AI infrastructure startup announced a $14 million Series B on October 27, led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC and Micromax participating. The company says it will spend the new capital on its inference engine and orchestration platform, and on expanding cloud partnerships, with North America a priority.
Nvidia backs software that tunes GPU workloads
Simplismart builds software for deploying and operating AI models on customers’ existing computing infrastructure. Its platform adjusts serving decisions such as hardware use, precision, batching, caching and scheduling to balance response time, throughput and cost. Cloud providers can also use the software to offer managed inference services on their GPU capacity.
“Your workload should dictate your infrastructure. Not the other way around,” wrote Amritanshu Jain, Simplismart’s co-founder and CEO, in the company’s funding announcement. The company’s pitch is that serving systems should be configured around what an application needs, rather than optimized for one headline speed or price measure.
Simplismart says its customers include HeyGen, Higgsfield, InVideo, Shiprocket, Apna and Ema, with workloads spanning video, voice, multimodal and enterprise AI. In one example from the announcement, it says a production configuration for Ema delivered latency of 180 milliseconds at seven cents per million tokens, against the customer’s targets of roughly 200 milliseconds and nine cents. Those figures are the company’s account of a customer deployment, not an independent benchmark.
Growth claims come with a limited disclosure
Simplismart’s announcement gives the 14-fold growth rate but does not state the underlying revenue figure. In an interview with TechCrunch, Jain described the annualized run rate as eight figures; the publication said the figure reflects recent customer usage, not contracted recurring revenue, and excludes revenue passed through to infrastructure providers. The company has not disclosed its exact run rate.
TechCrunch reported that the financing values Simplismart at about $100 million, including the new capital. The company’s announcement does not state a valuation. TechCrunch also reported that Nvidia has worked with Simplismart on technical integrations, including around Nvidia’s Dynamo inference software, and has introduced the startup to potential customers and partners.
The Nvidia investment carries a commercial tension
Simplismart says it can help customers get more from existing GPU capacity, including by allocating workloads a fraction of a GPU rather than a whole one. That proposition can appear at odds with Nvidia’s position as a leading supplier of AI chips. Jain told TechCrunch that more efficient inference could make Nvidia hardware more attractive and encourage broader use of AI applications.
The startup competes not only with inference software from Nvidia but also with offerings from Amazon Web Services, Google Cloud and Microsoft Azure, according to TechCrunch. Its differentiation, Jain told the publication, is serving customers with strict service-level requirements and tight limits on total cost of ownership. The funding announcement does not disclose Nvidia’s investment amount or the terms of its participation.
From individual deployments to shifting GPU supply
Simplismart says its next technical challenge is matching demand with available capacity across regions and time zones. The company is building a scheduling layer intended to place workloads where capacity is available while respecting response-time requirements and customer rules on data residency and compliance. That would extend its work from tuning individual deployments to coordinating supply across cloud infrastructure.
The company says it already powers the serverless inference layer of Yotta’s Shakti Studio, where users can run models without managing the underlying infrastructure. Whether Simplismart can extend that model to more cloud providers—and match spare GPU capacity across locations without violating latency or data-residency limits—remains the concrete test of its next phase.