What does it take for a 10-person team to generate US$200M in annual recurring revenue? Or for a startup to reach US$250M ARR in under two years, without a massive fundraising round?

For Marc Manara, head of startups at OpenAI, the answer lies in a disciplined blend of speed, focus, and the intelligent use of AI infrastructure. Speaking at OpenAI x SCB 10X in Bangkok, Manara pulled back the curtain on how the company works with founders worldwide, and why its platform is becoming a launchpad for the next generation of market leaders.

From workflows to full-stack agents

Manara’s definition of an AI “agent” was intentionally functional: a workflow to guide behaviour, tools to expand capability, and guardrails to ensure appropriate, ethical outcomes. In his telling, agents are not novelties; they are the new operational backbone.

The real shift, he argued, will be from isolated AI functions to full-stack solutions capable of handling multi-step processes, integrating into existing workflows, and adapting across industries.

The 2025 investment lens

OpenAI’s priorities for the year ahead underscore this trajectory:

• Models and customisation
• An agents platform
• Multimodality

Multimodality, where text, audio, and images flow seamlessly through one system, is already reshaping product design. Manara framed it as a “first-class capability,” not an experimental feature.

Lean teams, outsized returns

Two examples dominated the discussion:

• Cursor: 20 employees, US$250M ARR in just 21 months
• Midjourney: 10 employees, US$200M ARR in two years

Both exemplify what Manara calls seedstrapping: building significant traction and revenue before pursuing large-scale funding. The model favours rapid iteration, tight feedback loops, and a relentless focus on product-market fit.

Also Read: AI gold rush: How OpenAI’s Singapore expansion could reshape the startup ecosystem

Beyond the API

OpenAI’s engagement with startups goes far beyond providing API access. Programmes include:

• Enhanced concierge support: solution architects, account escalation, and dedicated office hours
• Exclusive resources: API credits, invite-only technical sessions, and “build hours” with OpenAI experts
• Direct influence: alpha and beta access to shape the product roadmap

This is underpinned by OpenAI’s internal loop: Research → Apply → Deploy → Repeat, which Manara urged founders to replicate.

For those paying close attention

Not everything shared in Bangkok was on the slides. Manara pointed to two resources that, while technically public, are rarely promoted and often overlooked: one on the emerging frontier of text-to-speech, and another on a discreet pathway to privileged access within OpenAI’s startup ecosystem.

I will share both with readers who follow me here on e27. Think of it as a private briefing for those who are actively paying attention.

A global play, from Bangkok

The keynote underscored a bigger truth: global-scale AI infrastructure is no longer a Silicon Valley monopoly. As OpenAI expands its reach into Southeast Asia, the advantage will go to founders who can not only access these capabilities but operationalise them faster than their peers.

Want the two “golden nuggets” I mentioned? Follow me here on e27 and I’ll send them to you directly. Sometimes, knowing where to look is the real advantage.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image courtesy of the author.

The post From Bangkok to billions: Inside OpenAI’s startup growth playbook appeared first on e27.