Giga Raised $61M for Voice AI Agents — and Its Founders Did It While Fighting Racist Attacks

# Giga Raised $61M for Voice AI Agents — and Its Founders Did It While Fighting Racist Attacks

Two Indian engineers built a voice AI agent company in San Francisco, raised $61 million, and did it while publicly documenting the racist harassment they faced along the way. The story got attention for the wrong reasons. It deserves attention for the right ones.

Giga is not a feel-good story about resilience. It is a signal about where the AI agent economy is heading — and who is building the infrastructure that will run it.

What Giga actually does

Giga is a San Francisco-based startup building voice-based AI agents designed to handle customer support conversations at scale. We are not talking about chatbots that route tickets. Giga's agents conduct full voice conversations — hundreds of thousands of them, daily — without a human in the loop.

The target market is any company currently paying for a customer support operation. That is a large, expensive, and structurally inefficient market. According to IBISWorld, the US call center industry alone generates over $24 billion in annual revenue. Giga is not trying to improve that industry. It is trying to make most of it unnecessary.

The $61 million raise — confirmed in 2025 reporting — puts Giga in a small group of voice AI agent companies that have cleared the Series A threshold with a product already in production. That distinction matters. A lot of agent startups are raising on demos. Giga is raising on deployed infrastructure.

Why voice, why now

Text-based AI agents — chatbots, email automation, support widgets — have been around long enough that customers distrust them on sight. Voice is harder to fake badly. A voice agent that sounds natural and resolves issues on the first call does not feel like a downgrade. A clunky chatbot always does.

The technical gap that made high-quality voice AI impossible two years ago has closed. Latency is down. Speech synthesis is indistinguishable from human in many contexts. Large language models can now hold context across a multi-turn conversation without losing the thread. Those three things converging is what makes Giga's timing credible.

For founders building AI-native companies in 2026, voice is becoming a serious deployment surface — not a novelty. Giga is one of the clearest examples of a company treating voice agents as production infrastructure rather than a product feature.

The funding context

$61 million is a meaningful number in the current environment. AI infrastructure raises at this level are not rare, but they are not automatic either. Investors at this stage want evidence of unit economics, retention, and a defensible position against well-funded competitors.

Giga's defensibility argument appears to be data. Every conversation its agents handle generates training signal. The more calls the system processes, the better it gets at handling edge cases in specific industries. That creates a compounding moat that is difficult for a generic LLM wrapper to replicate.

This is the same logic behind every serious AI agent stack worth studying in 2025 and 2026: the asset is not the model, it is the feedback loop.

The racist attacks and why they matter to founders

Giga's founders went public about the harassment they received — racist comments directed at them as Indian engineers building in San Francisco. They did not stay quiet, and the story spread.

This is relevant to founders reading this site for one reason: the pipeline of world-class AI engineers is global, and a significant share of the most technically sophisticated AI startups being built right now have Indian founders. Giga is one. So are dozens of companies in the agent infrastructure layer that will determine what software looks like in five years.

The harassment the Giga founders faced is an attempt to make that pipeline feel unwelcome. It failed. They raised $61 million. The work continues.

What Giga's model implies for smaller operators

Giga is venture-backed and building for enterprise scale. But the architecture it represents — autonomous voice agents replacing headcount in customer operations — is already filtering down to smaller companies.

Sprinkal, an AI marketing agent team built by Julien de Waal, operates on a similar principle in a different vertical: deploy agents to handle work that previously required a team, measure the output, and optimize the loop. The underlying logic is the same whether you are handling customer support calls or running paid acquisition campaigns.

The question for any founder watching Giga is not "how do I build the next Giga?" It is "which part of my operation still runs on human labor that an agent could replace today?"

Revenue per employee as the real metric

Giga's raise is impressive. But the number that will define whether Giga is actually building something structurally different is revenue per employee — not ARR, not valuation.

If Giga is replacing customer support operations at scale while keeping its own headcount lean, its revenue-per-employee ratio should look nothing like a traditional SaaS company. That is the thesis. A company selling agent infrastructure that itself relies on large operational headcount is not proving the model — it is just selling picks and shovels while mining with a shovel.

For founders tracking where AI is genuinely compressing labor costs versus where it is adding overhead, revenue per employee is the number that cuts through the noise. Giga's next few years of hiring decisions will tell that story more clearly than any funding announcement.

What to watch

Three things will determine whether Giga becomes a case study in AI-native company building or a cautionary tale about over-capitalized infrastructure plays:

First, churn. Voice AI agents that fail on complex calls push customers back to human support. If Giga's agents cannot handle the long tail of difficult conversations, enterprise customers will not fully commit.

Second, gross margin. Inference costs at the scale Giga is targeting are real. The unit economics only work if the margin per conversation is high enough to justify the infrastructure.

Third, headcount discipline. The companies worth tracking on the one-person unicorn model are the ones that resist the temptation to hire their way through growth. With $61 million in the bank, that discipline is harder to maintain.

Giga is not a one-person company. But the model it is building — agents handling work at scale with minimal human intervention — is exactly the infrastructure that makes lean, high-revenue-per-employee companies possible. Whether Giga itself stays lean enough to prove that internally is the open question.

The founders built something real, raised real money, and kept building through conditions that would have stopped a lot of people. That part of the story is not complicated.

---

Is your company eligible? Submit to the leaderboard → onepersonunicorn.co/submit