Nvidia joins $14 million funding round for AI inference startup Simplismart

New funding for inference software

The Bengaluru-based startup Simplismart has secured $14 million in a Series B funding round. This investment was led by Dallas Venture Capital and included participation from major names like Nvidia, Accel, Shastra VC, and Micromax. The funding round values the four-year-old company at approximately $100 million.

The capital comes at a time of rapid growth for the company. Simplismart reported that its annualized revenue run rate has increased 14-fold over the last six months, reaching an eight-figure sum. This growth highlights a growing demand for specialized software that manages how AI models interact with hardware.

The startup plans to use this new capital to fuel its expansion into the United States market. They also intend to grow their current team of 58 employees and invest more heavily in their core software offerings. By scaling their operations, the company aims to reach a broader audience of enterprise customers.

How the software optimizes AI

Simplismart operates in the space of AI inference infrastructure. Inference is the process of running a trained AI model to generate an output, such as text or images. The company builds software that sits between the AI models and the physical computing hardware they run on.

The platform tunes how these models are deployed to help companies find a balance between three competing factors. These factors include the speed of the response, the volume of requests the system can handle, and the amount of compute power required. It provides tools for fine-tuning models, scaling compute automatically, and monitoring performance in real-time.

Unlike some inference providers, Simplismart does not own its own GPU (Graphics Processing Unit) capacity. Instead of buying and reselling hardware, it focuses on the software layer. This allows the company to provide optimization tools that work across different types of infrastructure without the overhead of managing physical data centers.

Cost savings for major customers

The primary goal of Simplismart's technology is to make AI more affordable for businesses. The company claims its software can help customers significantly reduce their operational expenses. For example, it helped the company InVideo cut its GPU serving costs by 56%.

Other clients have seen similar results in reducing the cost of running their specific applications. The communications platform Exotel reported a reduction in cost per minute of up to 57% after using the technology. These numbers demonstrate how software optimization can directly impact a company's bottom line when deploying AI.

The startup serves a diverse range of clients, including Higgsfield, Ema, and the Indian cloud provider Yotta. Yotta specifically uses the technology for the serverless AI inference service on its Shakti Studio platform. These partnerships show that the software is being adopted by both global tech firms and regional cloud providers.

Strategic partnership with Nvidia

Nvidia's involvement in the funding round is significant because of its role as a leader in AI hardware. Beyond just providing capital, Nvidia is working directly with Simplismart on technical integrations. This includes work specifically around Nvidia's Dynamo inference software.

Nvidia has also played a role in helping the startup connect with new opportunities. The hardware giant has assisted in introducing Simplismart to potential new customers and partners. This collaboration suggests that hardware manufacturers are increasingly interested in software that helps their chips run more efficiently.

The partnership reflects a broader trend in the industry where hardware and software must work in harmony. By integrating with Nvidia's tools, Simplismart can ensure its optimization software works seamlessly with the most common AI chips. This helps customers get the best performance out of their existing hardware investments.

The economics of AI inference

The startup's leadership emphasizes that there is no single best way to run AI models. Co-founder and CEO Amritanshu Jain noted that "Nobody actually needs the fastest or the cheapest inference." Instead, he explained that customers need speeds and costs that make sense for their specific applications.

This perspective highlights the shift toward