正文 · AI 翻译
IT Home reported on October 28 that, as the artificial intelligence boom continues to strain computing resources, Nvidia participated in a $14 million funding round for Indian startup Simplismart (IT Home note: approximately 93.573 million yuan at current exchange rates). The investment targets software that can make AI models more efficient to run and, in some cases, help companies use fewer GPUs.
The round was a Series B consisting entirely of newly issued equity. Alongside Nvidia, Accel, Shastra VC and Micromax invested, with Dallas Venture Capital leading the round. Following the deal, the startup, founded just four years ago, was valued at approximately $100 million (about 668 million yuan at current exchange rates), including the funds raised in this round.
Companies are now spending heavily on GPUs to run AI applications. But making these models both reliable and economical when deployed at scale remains a significant challenge. Different workloads have different requirements for speed, cost and performance. To meet business needs, companies often have to make complex parameter adjustments when deploying and operating AI models.
Simplismart says its software can automate much of this work, adjusting how a model runs on existing computing infrastructure according to its actual needs, without requiring customers to manage these complex configurations themselves.
Simplismart co-founder and CEO Amritanshu Jain (pictured on the left above) said in an interview, “In reality, nobody truly needs the fastest or cheapest inference solution.” He explained that companies need a solution suited to their particular use case, allowing AI models to run at a reasonable speed and cost.
Simplismart’s software sits between AI models and the underlying computing infrastructure, adjusting how models are deployed and managed to meet customers’ specific needs. The company says its software can optimize metrics including model response speed, request throughput and computing resource consumption, helping companies make fuller use of their existing GPU resources.
The Bengaluru, India-based startup offers companies several ways to deploy and run AI models: through Simplismart’s own platform, on customers’ existing cloud infrastructure, or in on-premises environments.
The company also provides model fine-tuning tools, can automatically scale computing resources up or down as needs change, and monitors model performance. Cloud providers can also use Simplismart’s software to build and operate their own AI inference services on GPU infrastructure.
Simplismart says its software helped AI video generation platform InVideo reduce GPU service costs by 56%. The company also works with AI video companies HeyGen and Higgsfield, as well as enterprise AI startup Ema, among others. Indian cloud provider Yotta also uses Simplismart’s technology in its Shakti Studio platform to support serverless AI inference services.
Jain told TechCrunch that Simplismart’s annualized revenue run rate grew 14-fold over the past six months, reaching eight figures. However, the metric is extrapolated from customers’ recent actual usage rather than contracted recurring revenue, and excludes revenue passed through to infrastructure providers. The company did not disclose a specific amount.
Jain said that although Simplismart has more customers in India, more than half of its revenue comes from the U.S. market.
Alongside its investment in Simplismart, Nvidia is working with the company on technical integrations, including integration involving Nvidia’s Dynamo inference software. Jain told TechCrunch that the collaboration involves Simplismart engineers and Nvidia solutions architects. Nvidia is also helping Simplismart connect with potential customers and partners.
Notably, the investment appears to present a contradiction: as the dominant supplier of AI chips, Nvidia is backing software that can help companies reduce some of their GPU usage.
Jain believes more efficient AI inference could ultimately benefit Nvidia as well. More efficient inference could make its hardware more attractive to customers and encourage the deployment of more AI applications, expanding overall market demand.
The AI inference software market in which Simplismart operates is becoming increasingly crowded. Its competitors include solutions from Nvidia itself, as well as major cloud providers and hyperscalers such as Amazon Web Services, Google Cloud and Microsoft Azure.
As demand for AI applications grows, companies are increasingly seeking to make better use of expensive GPU resources. More startups are therefore developing tools intended to make AI models run faster and at lower cost.
Jain said, “When customers have stricter requirements for service-level agreements (SLAs) and total cost of ownership (TCO), we often have an advantage.” This means customers require systems to meet strict performance standards while also controlling overall costs.
He said that, unlike some competitors that offer preset solutions for high-speed or low-cost inference, Simplismart aims to find a more suitable balance among speed, cost and performance based on each customer’s specific needs.
Simplismart was founded in 2022 by Jain, a former Oracle engineer, and Devansh Ghatak (pictured on the right above), a former Google engineer. The two met while studying at BITS Pilani in India and later founded the company based on their experience building and deploying machine learning systems.
Jain said Simplismart currently has 58 employees.
Simplismart plans to use the proceeds from this round to expand further in the U.S. market, grow its team and continue investing in AI inference software research and development. Jain told TechCrunch that the U.S. remains a key market for the company and that Simplismart hopes to expand its business there further on the basis of its existing customer base.
Before the funding announcement, Indian English-language newspaper The Economic Times reported that Simplismart was in talks to raise $40 million to $50 million (approximately 267 million to 334 million yuan at current exchange rates) at a target valuation of $300 million to $400 million (approximately 2.005 billion to 2.674 billion yuan at current exchange rates).
Jain declined to comment on that report and said the company was not actively seeking another funding round. He also emphasized that the Series B round had closed and would not be increased.