Recommended Articles
Fundraising

Simplismart Raises $14M Series B to Make AI Inference Cheaper and Faster

By Angeled · November 1, 2026 · Fundraising
Simplismart Raises $14M Series B to Make AI Inference Cheaper and Faster
Simplismart Raises $14M Series B to Make AI Inference Cheaper and Faster

Simplismart, a Bengaluru-based AI infrastructure startup, announced on October 27, 2026 that it has raised $14 million in a Series B funding round led by Dallas Venture Capital, with participation from NVIDIA, Accel, Shastra VC, and Micromax. Dallas Venture Capital invested close to $5 million in the round. The funding follows a $7 million Series A led by Accel in October 2024.

The company plans to use the capital to strengthen its AI inference platform, expand partnerships with cloud providers and sovereign AI players, and grow in North America and other international markets. It is also building a dynamic GPU scheduling layer that matches AI workloads with available computing capacity across regions and time zones.

Simplismart was founded in 2022 by Amritanshu Jain and Devansh Ghatak, both alumni of BITS Pilani. The company says its annualized revenue run rate grew 14-fold over the past six months, and around 55 percent of its revenue comes from the United States, with the remainder generated in India.

What inference is, and why it costs so much

Training an AI model — teaching it from vast amounts of data — gets most of the headlines, but inference is where most of the money is spent. Inference is the process of running a trained model to generate outputs: answering a question, transcribing audio, generating a video, or summarizing a document. Every time a user interacts with an AI product, inference happens, and it consumes computing power on graphics processing units, or GPUs.

GPUs are expensive and scarce. A single high-end AI chip can cost tens of thousands of dollars, and running large models at scale requires fleets of them, along with power, cooling, and networking. For companies whose products depend on AI, inference costs are often the largest line item after salaries — and they scale with usage, meaning success brings higher bills.

That is the problem Simplismart attacks. Its software sits between AI models and GPU infrastructure, managing deployment, scaling, and operation so that each workload runs at the right balance of latency, throughput, quality, and cost. Its modular inference engine searches for the best serving configuration for each workload rather than applying a one-size-fits-all setup.

A concrete example from the company illustrates the point. For customer Ema, an enterprise AI company, the target was roughly 200 milliseconds of latency at about 9 cents per million tokens. Simplismart's optimization search surfaced configurations at 5 cents per million tokens at 350 milliseconds and 12 cents at 100 milliseconds, then delivered a production configuration at 7 cents and 180 milliseconds — meeting both the cost and latency requirements. Small savings per request compound enormously at production scale.

A software layer, not a GPU landlord

Simplismart's business model differs from many inference providers. It does not buy or reserve GPU capacity and resell computing services. Instead, it supplies the software layer that lets third-party GPU infrastructure run production AI workloads efficiently. That positions it as an enabler for data center operators and cloud providers rather than a competitor to them.

This distinction matters strategically. Cloud providers and sovereign AI programs — governments building national AI infrastructure — need software to turn raw GPU fleets into services they can sell. Simplismart works with NVIDIA to help providers deploy and manage AI infrastructure at scale, and its platform powers the serverless inference layer of Yotta's Shakti Studio platform in India, letting customers run models without managing the underlying infrastructure.

Its customer list spans video, voice, multimodal, and enterprise AI workloads, including HeyGen, Higgsfield, InVideo, Shiprocket, Apna, and Ema. The mix of U.S. and Indian revenue, plus the NVIDIA relationship, suggests a company deliberately building across both markets rather than depending on one.

Why inference economics are the next battleground

As AI models become cheaper to access and more widely deployed, the competitive edge shifts from who has the best model to who can run models most efficiently. Every percentage point of inference cost saved flows directly to margins for AI application companies, and every improvement in GPU utilization defers the next expensive hardware purchase for cloud providers.

That is why Dallas Venture Capital's managing partner Dayakar Puskoor described inferencing as one of the biggest game changers in AI economics. The investors in this round — a U.S. venture firm, the world's leading AI chipmaker, a top-tier global fund in Accel, and Indian venture capital — reflect the global nature of the opportunity. Inference optimization is not a regional story; it is a cost problem every AI company on earth shares.

Why it matters for founders and investors

For founders building AI products, Simplismart's round is a prompt to audit your inference bill before your next fundraise. Model choice, serving configuration, and infrastructure partnerships can move unit economics more than a new feature. If inference costs grow faster than revenue, the business model breaks no matter how good the product is.

For investors, the round highlights a maturing AI stack. The first wave of AI funding went to foundation models; the current wave is going to the infrastructure that makes those models affordable to run. Companies that sit between models and hardware — optimizing, orchestrating, and monetizing GPU capacity — are becoming a distinct and defensible category, especially when they win both enterprise customers and cloud-provider partnerships.

Angeled.com connects founders with investors.

Source: inc42.com
Source: aithority.com
Source: globalventuring.com

Read Next