The rise of generative and agentic AI has redefined what it means to start a company. Capital, talent, and timing still matter, but in today’s ecosystem, so does the architecture that underpins innovation. The AI tech stack is no longer a commodity layer—it is the competitive moat.
Every major technological shift has created a new kind of entrepreneur. The industrial age rewarded those who mastered machines; the internet favoured those who could code at scale. The AI era is shaping founders who can build directly on intelligence itself. Darren Mowry, Vice President of Global Startups at Google Cloud, has witnessed this transformation unfold in real time. “AI has accelerated startup innovation more than any technology since perhaps the internet itself,” he explains. “We have a front-row seat to the way this technology is fuelling entirely new industries.”
That front-row seat provides a unique perspective on how the startup landscape is reorganising around AI infrastructure. “Nine of the top ten AI research labs use Google Cloud, as does nearly every AI unicorn,” Mowry notes. “More than sixty per cent of the world’s generative AI startups are now building with our technology stack, and the trend is accelerating.”
The attraction, he says, is not brand loyalty but technical architecture. “Founders are choosing environments that give them both speed and choice,” he continues. “They need to move from concept to production faster than ever before, and they need infrastructure that lets them train, scale, and serve AI models efficiently, without sacrificing flexibility.”
Over the past year, Google Cloud has seen a twenty-per-cent increase in the number of new AI startups, those raising Series A or B rounds, joining its ecosystem. “These are the founders shaping the next wave,” Mowry adds. “They are not simply optimising existing products; they are building entirely new business models that depend on AI at the core.”
This dependency changes the physics of innovation. Where once startups had to build infrastructure from scratch, they now assemble intelligence. Compute power, data orchestration, model training, and deployment are all part of a composable architecture. The infrastructure is becoming the innovation layer itself.
Speed and scale become inseparable
AI has compressed the distance between an idea and its execution. Startups that once needed years to mature can now reach product-market fit in months. “We have seen founders go from whiteboard concept to scaled deployment in weeks,” Mowry says. “That speed is possible because AI development has become modular and accessible. Startups can stand on the shoulders of hyperscale infrastructure and focus their energy on differentiation rather than plumbing.”
He describes this as the new equilibrium between speed and scale. “What used to be a trade-off, moving fast or building reliably, is now a convergence,” he explains. “The right AI stack enables both. You can experiment quickly and scale globally without rewriting the foundation.”
Mowry highlights how AI startups are taking advantage of these capabilities to solve increasingly specialised problems. “We are working with founders across every vertical, healthcare, media, education, and cybersecurity, who are using generative and multimodal AI to transform their industries,” he says. “They are not just making existing systems smarter; they are inventing new categories of product.”
He points to examples such as healthcare innovators like CerebraAI, which is fine-tuning medical imaging models to detect disease earlier and more accurately, and creative platforms like Krea.ai, which use generative video models to redefine how content is produced. “The diversity of use cases is staggering,” Mowry adds. “But the common thread is the same, each of these companies relies on a solid technical foundation to keep pace with its own innovation.”
The most interesting development, he argues, is the rise of AI-native businesses. “These are companies that are AI all the way down,” he explains. “AI is not a feature of their product; it is their product. The intelligence layer defines their operations, their customer engagement, and their revenue model. That is a profound shift in how startups are built.”
The architecture of orchestration
Behind this transformation lies a change in how infrastructure itself is conceived. “The modern AI stack is designed for orchestration,” Mowry says. “We are moving from an era of isolated compute to one of connected intelligence, where AI acts both as workload and orchestrator.”
He describes how AI is beginning to manage its own ecosystem. “Agentic systems can already decide which tasks should run on CPUs, GPUs, or TPUs, and soon they will determine how to route and optimise entire workflows,” he explains. “This is what we call ‘AI managing AI’, the orchestration of intelligence across hardware, models, and data pipelines.”
To enable this, startups need unified platforms that simplify complexity. “Developers are no longer locked into one environment,” Mowry continues. “They can move seamlessly between chips, model families, and frameworks, deploying across multiple regions without friction. That flexibility is critical in a world where the underlying technology evolves every quarter.”
He believes this orchestration mindset will become the defining skill for modern founders. “The next generation of entrepreneurs will not just code, they will compose,” he says. “They will build systems that learn how to allocate resources, scale automatically, and optimise performance without direct human intervention. The real innovation is not in writing code, but in teaching systems how to improve themselves.”
This new model of computing changes the rhythm of startup growth. “In traditional software, scale was linear,” Mowry explains. “You added more servers, more engineers, more users. In AI-driven architectures, scale becomes exponential because learning compounds. The more data you feed into the system, the better it becomes. The startup that masters orchestration is the one that compounds the fastest.”
AI-native businesses reshape startup logic
This shift is already changing how investors value young companies. “We are starting to see metrics evolve beyond revenue or user acquisition,” Mowry says. “Investors are now asking about model efficiency, inference cost, and data diversity. They want to know how adaptable the startup’s architecture is, how quickly it can pivot to new models or integrate new modalities.”
He believes this reflects a broader maturity in how the market understands AI. “The hype phase is fading, and what replaces it is substance,” he continues. “Founders who can explain their data strategy, their model lineage, and their orchestration logic are the ones attracting serious investment.”
Mowry also acknowledges that the shift to AI-native business brings responsibility. “Startups cannot ignore the governance side,” he warns. “As AI becomes embedded in decision-making, explainability, fairness, and security are no longer optional. The startups that build trust into their systems from day one will have an enduring advantage.”
He sees an encouraging trend in the way early-stage founders approach these issues. “There is a new generation of entrepreneurs who see responsible AI not as a compliance hurdle but as a design principle,” he says. “They understand that ethical deployment is a prerequisite for scale. Customers and regulators alike are demanding it.”
The global dimension of this transformation is equally striking. “Innovation is not concentrated in one geography anymore,” Mowry observes. “We are seeing extraordinary AI startups emerge from India, Europe, the Middle East, and Africa. They are solving local problems with global implications, from logistics optimisation to multilingual communication. Access to scalable infrastructure has democratised where innovation happens.”
The orchestration mindset defines advantage
For Mowry, the next phase of AI entrepreneurship will be defined by orchestration rather than invention. “The biggest opportunities lie in how you connect technologies, not just in how you create them,” he says. “Every startup now has access to powerful models, chips, and data pipelines. The differentiator is how intelligently those components are combined into a working system.”
He believes that orchestration will become the new competitive advantage. “Two companies might use the same foundation model,” he explains. “But the one that orchestrates its data flow better, that optimises inference across hardware, and that automates scaling intelligently, that company will win. Architecture will decide outcomes.”
Mowry’s advice to founders is deceptively simple. “Think like an architect, not a tenant,” he says. “Do not treat the AI stack as a fixed service; treat it as a canvas. Every choice, model, framework, or chip, has strategic implications. The founders who understand that will move faster, spend less, and deliver more.”
This perspective reframes the relationship between startups and hyperscale providers. “We see ourselves as enablers,” Mowry adds. “Our role is to make sure startups have the freedom to experiment, to access world-class compute, and to build responsibly. But the real innovation always happens at the edges, where founders take these tools and create something entirely new.”
He emphasises that the startup ecosystem has never been more dynamic. “There has never been a better time to build,” he concludes. “The cost of experimentation is lower, the tools are smarter, and the market appetite for AI-driven products is immense. The constraint is no longer access to technology, it is imagination.”
AI has compressed the timeline of innovation to near real time. Founders are now operating in a world where models evolve weekly, infrastructure adapts automatically, and the boundary between human and machine creativity continues to dissolve. The companies that thrive will be those that view AI not as an accessory but as architecture—a foundation that learns, scales, and competes alongside them.
The future of startups will not hinge on who raises the most capital or deploys the largest model. It will belong to those who understand that the stack is the strategy, that in an era of intelligent infrastructure, orchestration is everything.
