Simplismart Raises $14 Mn Series B Led By Dallas Venture Capital, NVIDIA Joins

Simplismart, a Bengaluru-based AI inference infrastructure startup, has raised $14 million in a Series B round led by Dallas Venture Capital. NVIDIA, Accel, Shastra VC and Micromax participated, along with angel investors.

The round was entirely primary, meaning all the money is fresh capital, and it values the four-year-old company at about $100 million post-money. Total funding now stands at roughly $21 million, after a $7 million Series A in October 2024.

What Simplismart Actually Sells

Inference is the stage where a trained AI model is put to work. It answers a query, generates a video clip, or produces a voice. It is also where most AI budgets quietly die, because running a model costs money every time someone uses it.

Read More: Ex-Slice Executive’s QwikBuild Raises $2.3 Mn, Plans Japan Joint Venture With Globalway

Start UpSarvam Launches Prism, A 44-Millisecond Decision Model On Sarvam 105BFundingFal Raises $1.3 Bn At $16 Bn Valuation As AI Inference Demand Outruns Capacity

Simplismart builds the software layer between AI models and the GPUs they run on. Its inference engine decides how a model should run based on latency, throughput, quality and cost. Its orchestration platform handles deployment and scaling across cloud and on-premises infrastructure.

Read More: Shirley Setia’s ISLAND Beauty Raises Rs 2.58 Cr From AJ VC

The company describes the interface as declarative and Terraform-like: a team states what it wants, and the platform configures the serving stack to deliver it. That control plane works across clouds, regions and on-premise gear, so the same cost and compliance rules follow a workload wherever it runs.

The Numbers It Is Selling

Simplismart says its annualised revenue run rate has grown 14-fold over the past six months. Customers include HeyGen, Higgsfield, InVideo, Shiprocket, Apna and Ema, running production workloads across video, voice and multimodal AI.

The clearest proof point is Ema, an enterprise AI company. It wanted roughly 200 milliseconds of latency at about 9 cents per million tokens. Simplismart’s optimisation search found two options: 5 cents at 350 ms and 12 cents at 100 ms. It then shipped a production setting at 7 cents and 180 ms.

It also claims it cut video platform InVideo’s GPU serving costs by 56%. Cost reductions of that size are the reason inference startups exist. A model that is 50% cheaper to run changes the unit economics of an entire product.

Why NVIDIA Wrote a Cheque

NVIDIA’s participation is the signal investors will read. Simplismart is an early member of the NVIDIA Inception programme and has worked directly on integrations around NVIDIA’s inference software, including Dynamo.

There is an obvious tension in that relationship. Simplismart’s software helps companies do more with fewer GPUs, and NVIDIA sells GPUs. Amritanshu Jain, co-founder and chief executive, argues the two goals are aligned. Cheaper inference means more AI applications get built, he says, which means more hardware is eventually sold.

NVIDIA has also introduced the startup to customers and partners, according to the company. For a young infrastructure business, those introductions are worth as much as the money.

The Cloud-Provider Bet

Unlike inference providers that buy or reserve GPU capacity and resell it, Simplismart does not own computing infrastructure. Its software lets cloud providers turn existing GPU fleets into managed AI services. Those are then sold under the provider’s own brand.

Yotta is one such partner in India. Simplismart powers the serverless inference layer of Yotta’s Shakti Studio, which lets customers run models without managing the hardware underneath.

That model is attractive to sovereign AI programmes, where governments and local providers want AI capacity that stays inside their jurisdiction. Simplismart is building towards a dynamic GPU scheduling layer, so idle capacity in one region can serve demand in another when requirements allow.

What to Watch

Three things will decide whether Simplismart becomes infrastructure or a feature. The first is retention. Inference customers can switch providers in weeks, and 14x growth in six months says nothing about whether they stay.

The second is the cloud-partner pipeline, which the company says spans India, North America, the Middle East, Australia and East Asia. Each deal converts GPU capacity into recurring software revenue, but only if the providers actually sell it.

The third is competition from above. NVIDIA, AWS, Google Cloud and Microsoft Azure all sell inference. They also own the hardware relationship Simplismart depends on. The startup’s defence is neutrality across clouds, which is a real advantage until one of those platforms decides to be cheaper.