The first thing you notice about Upscale Studio is that it looks less like a media-buying console and more like a creative tool. The product, unveiled in May, lets a brand upload its assets and walk out the other side with a finished Streaming TV spot, then bid on inventory programmatically against performance goals like new-customer acquisition or site visits [BusinessWire, May 2025]. The implicit pitch to a direct-to-consumer marketer is unmistakable: treat a 30-second CTV placement the way you already treat a Meta ad, with the creative generated, the buy automated, and the optimization continuous.
That is the wager Upscale AI is making, and it is a specific one. The San Francisco company, founded in 2020 by Amit Mastud, Sandeep Singh, and Yeshu Singh, came out of stealth in May 2025 with $5.5 million in seed funding led by NVP Capital, with participation from Microsoft's M12, Eniac Ventures, SuperAngel.Fund, and Breakpoint Capital [BusinessWire, May 2025]. The product, branded Upscale Studio, is positioned as the first end-to-end AI-native platform built specifically to create and run performance ads on Streaming TV [Upscale AI, retrieved 2025]. Two halves, fused: generative video on the front end, machine-learning programmatic bidding on the back.
The bet
The wedge is the gap between how brand TV and performance digital have always been bought. Television has historically been a reach medium, sold in upfronts, measured in gross rating points, and creatively expensive enough that an early-stage commerce brand simply could not afford to test it. Streaming changed the inventory side of that equation, but the creative side stayed stubbornly bespoke. Upscale's argument is that if you can compress the cost of producing a competent CTV spot to something close to the cost of a static social ad, and you can bid on the impression in real time against a conversion goal, then the whole funnel-bottom playbook that lives on Meta and Google can finally migrate to the living room. The company describes its bidding stack as state-of-the-art ML models running over real-time programmatic auctions, optimized against goals like acquisition and site visits [Upscale AI, retrieved 2025].
Why it could be big
The tailwinds here are not subtle. Connected TV ad spend has been the fastest-growing line item in most digital media plans for several years, and the supply side, from Netflix's ad tier to Amazon Prime Video to the FAST ecosystem, has multiplied the available inventory faster than most performance marketers know how to deploy against it. The infrastructure for buying it programmatically exists. What has lagged is the creative pipeline, because making a watchable TV ad has historically required an agency, a shoot, and a six-figure budget. A platform that collapses that into a software workflow has a credible shot at becoming the default on-ramp for the next thousand commerce brands trying CTV for the first time.
The investor syndicate reflects that thesis. NVP Capital led; M12 brings Microsoft's adtech and cloud surface area; Eniac has a long history with consumer and adtech infrastructure bets [BusinessWire, May 2025]. None of these are tourist checks in the category.
The team and traction
| Round | Date | Amount | Lead |
|---|---|---|---|
| Seed | May 2025 | $5.5M | NVP Capital |
Mastud, Singh, and Singh built the company with a founding group drawn from Google, Twitter, Salesforce, Moloco, and MNTN [BusinessWire, May 2025]. That last name matters more than the others for context: MNTN has spent the last several years arguing publicly that performance TV is a real category, and pulling talent from inside that operation suggests Upscale's founders understand the buyer, the measurement debate, and the sales motion required to land mid-market commerce brands. The company is currently hiring a New Business Sales lead in San Francisco who will own first-touch conversations with brands, sign initial insertion orders, and shepherd them through onboarding and creative production [Upscale AI Careers, retrieved 2025], a job description that reads like a company moving from beta into paid pilots.
The honest counterfactual
The most credible bear case is competitive density. MNTN, the same company several Upscale founders came from, has built a public brand around performance CTV, and the broader programmatic stack includes The Trade Desk, Roku's OneView, and Amazon's DSP, all of which can in principle bolt generative creative onto an existing buyer base. The question for Upscale is whether an AI-native workflow built from scratch beats a generative feature shipped on top of an incumbent's distribution. The bull answer, supported by the product positioning, is that the creative-to-bidding loop is the actual product, not a feature: a brand that uploads assets at 9 a.m. and is bidding on impressions by lunch is buying something that an incumbent stitched together from acquisitions cannot easily replicate [Upscale AI, retrieved 2025]. Whether that holds up under contact with enterprise procurement is the thing to watch.
What to watch
The next twelve months should answer the most important questions. Look for the first set of named brand customers, particularly in DTC commerce verticals where ROAS is measured weekly. Watch for a Series A in the back half of 2026, which the seed syndicate and the M12 relationship would naturally support if pilot economics hold. And watch the creative output itself: the credibility of an AI-generated CTV ad is ultimately a question a viewer answers in the first three seconds, not a metric a dashboard reports.
The cultural question Upscale is implicitly answering is whether television, the last advertising medium that still felt handmade, is ready to be bought the way a search keyword is bought. The living room is about to find out.