Simplismart raises $14 million Series B with Nvidia backing
Bengaluru AI inference startup Simplismart raised $14 million led by Dallas Venture Capital, with Nvidia and Accel in, at a reported $100 million valuation.
Simplismart, a Bengaluru startup that makes software for running AI models cheaply, raised $14 million (about Rs 133 crore) in a Series B round led by Dallas Venture Capital, Entrackr and Inc42 reported on 27 October 2026. Nvidia, Accel, Shastra VC and Micromax also invested. An earlier Entrackr report put the valuation at about $100 million after the money went in. The round matters because it is an Indian company selling to the "inference" stage of AI, the daily running of models, and because it is one of the few Indian AI infrastructure rounds with Nvidia on the cap table.
What was announced
According to Entrackr and Inc42, the $14 million round was led by Dallas Venture Capital, which put in close to $5 million. Nvidia, Accel (which led the previous round), Shastra VC, Micromax and some angel investors took part. The previous round was a $7 million Series A led by Accel in October 2024, so the company has raised $21 million across those two rounds, a sum we have added up ourselves from the two reported figures.
The company will use the money to build out its inference engine and orchestration platform, a dynamic GPU scheduling layer, partnerships with cloud providers and sovereign AI programmes, and expansion in North America and other markets. Founder Amritanshu Jain is the chief executive. Inc42 says the company was founded in 2022 by Jain and Devansh Ghatak, the chief technology officer.
Two figures come from the company alone: its annualised revenue run rate has grown 14 times over the past six months, and about 55% of revenue comes from the United States with the rest from India. No base figure was disclosed for the run rate, so "14 times" cannot be turned into rupees. A multiple on an undisclosed base can be large when the starting point is small.
What inference is, and why it is a business
Training an AI model is like writing a textbook: it costs a lot and happens once in a while. Inference is every time someone uses the model, like every student reading a page. For a company running a chatbot, a video generator or a voice agent for thousands of users, inference is the recurring bill, and most of it goes to graphics processors (GPUs), the expensive chips that run these models.
Simplismart sits between the model and the chips. According to Entrackr, its platform helps companies deploy and run AI models while managing inference cost, delay (latency) and how busy the GPUs are. It lets cloud providers turn spare GPU capacity into AI services for business customers. Unlike some rivals, it says it focuses on the software layer rather than buying or reselling GPU capacity.
Inc42 describes three ways it sells. Shared endpoints are billed by tokens used (a token is a small chunk of text a model reads or writes) with no minimum fee. Dedicated endpoints combine rented GPUs with its software. And "bring your own compute" deployments suit regulated sectors such as banking, financial services and insurance, where data may need to stay inside a company's own systems.
Customers, partners and the valuation question
Entrackr lists customers including HeyGen, Higgsfield, InVideo, Shiprocket, Apna and Ema, using it for video, voice, multimodal and enterprise AI. At Yotta, an Indian data-centre company, Simplismart's platform powers the serverless inference layer of Shakti Studio. With Nvidia, Inc42 says, it helps cloud providers offer managed AI services under their own brands. The company says it works with cloud providers and sovereign AI programmes in India, North America, the Middle East, Australia and East Asia.
Secondary coverage listed in search results carried other claims, such as a reported cost saving on GPUs and earlier reports of valuation talks at a much higher level. We did not open those pages, and we treat them as unconfirmed. We also could not open YourStory's report, which was blocked. What can be said is that the $100 million valuation comes from Entrackr's earlier report and is not a number the company has stated in the pages we read.
Why would Nvidia invest? Nvidia has not explained its investment publicly, according to the coverage we reviewed. A reasonable reading is that software that makes GPUs more useful helps sell more GPU capacity to cloud providers. That is our interpretation, not a statement from the company or from Nvidia.
The competition and the market it is in
Simplismart competes with other inference providers, including large cloud platforms and specialist companies that rent GPUs and sell access to models. The size of the market is hard to pin down, and we have no verified figure to quote. Inc42 notes that the round comes amid a wave of investment in India's AI infrastructure, citing Tata Communications' plan to invest $400 million to $450 million in AI connectivity.
The business has a structural risk. Inference prices have tended to fall as chips and software improve, which squeezes any company selling "cost savings". It also depends on large partners: Nvidia as an investor and chip supplier and cloud providers as channel partners. Choosing the software layer rather than owning chips reduces capital needs but makes it easier for bigger players to copy features. These points are our analysis, not claims from the sources.
What it means for founders, buyers and investors
For Indian founders, the round shows that infrastructure software, not only consumer apps, can attract international investors, including at a modest valuation by US standards. A $100 million valuation on $14 million raised means the new investors own about 14% in total, and the Dallas VC share of close to $5 million is about 5%. These shares are our arithmetic on reported figures.
For companies buying AI services, vendors like this compete on cost per token and delay. Buyers should ask how a quoted saving was measured, on which model and at what volume, since claims depend on workload. For regulated firms such as banks, the bring-your-own-compute model is the one to examine, since it keeps data in-house.
Use the simulator to see how a round of this size dilutes existing owners. The defaults are the reported round size and valuation; the stake before the round is an example.
What to watch next
Watch for the company's revenue base in future disclosures, since 14 times growth needs a starting number to judge. Also watch whether more Indian inference startups raise, whether Nvidia's participation turns into joint offerings with cloud providers, and whether regulated Indian firms adopt bring-your-own-compute deployments. Simplismart said it is expanding in North America, so announcements of new cloud partners are the next markers.
Key takeaways
- Simplismart raised $14 million in a Series B led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC and Micromax participating, reported on 27 October 2026.
- The valuation is reported at about $100 million, from an earlier Entrackr report, not stated by the company in the coverage reviewed.
- Annualised revenue run rate is up 14 times in six months, but the base was not disclosed, and about 55% of revenue is from the United States.
- The product is software for running AI models cheaply, billed by token, on dedicated GPUs, or on a customer's own hardware.
- The main risks are falling inference prices and dependence on large chip and cloud partners.
Sources
- [Entrackr: Simplismart raises $14 Mn from Dallas Venture Capital, NVIDIA and others](https://entrackr.com/news/simplismart-raises-14-mn-from-dallas-venture-capital-nvidia-and-others-12758191)
- [Inc42: Simplismart bags $14 Mn to expand AI inference platform](https://inc42.com/buzz/simplismart-bags-14-mn-to-expand-ai-inference-platform/)
Tags: Startup funding AI funding AI inference Nvidia Startups