Scaling at the speed of purpose

Scaling AI with Purpose: Reinventing Workflows and Infrastructure for Success

Moderated by Mr. Jeff Wilser, the panel addressed the critical challenge of successfully scaling AI. Despite significant hype, many companies struggle to implement AI effectively and achieve a high return on investment, highlighting a gap between aspiration and practical execution.

Ms. Hayden Brown, CEO of Upwork, emphasized that simply retrofitting existing systems with AI is insufficient. True success demands a wholesale reinvention of workflows, systems, and talent. Upwork itself committed to building an AI-native experience, aiming for an end-to-end solution by 2026, focusing on creating substantial, not just incremental, value.

Ms. Ami Badani, CMO of Arm Ltd., concurred, stressing the need to reinvent the entire workload from software to infrastructure. Hyperscalers are optimizing the full stack for efficiency, recognizing power consumption as a critical rate limiter. Maximizing “intelligence per watt” is essential for driving ROI in AI applications.

Ms. Badani highlighted AI’s broad impact across Arm’s 325 billion devices. Mobile manufacturers are designing AI-first handsets for on-device processing, while autonomous vehicles and robotics optimize AI workloads. Data centers prioritize power efficiency for training. She predicts AI will be ubiquitous within 5-10 years, transforming nearly every market segment.

Ms. Brown identified two talent trends: the emergence of entirely new products and workflows, and the fractionalization of full-time work. Companies are increasingly leveraging freelance talent to acquire new skills quickly. This shift towards a dynamic, flexible workforce enables businesses to adapt rapidly to evolving AI opportunities and new roles like “GTM engineer.”

Ms. Badani pointed to mindset shifts and power as primary challenges. Companies must embrace experimentation. Fundamentally, power consumption limits exponential AI scaling, necessitating investments in power-efficient compute infrastructure and renewable energy sources for future data centers and AI agents.

Ms. Brown explained how product innovation is changing. R&D teams are now closely integrated with core product teams. Products are shipped earlier, often as prototypes, to gather crucial human feedback from customers. This iterative approach accelerates tuning and significantly improves overall product quality.

The discussion touched on “vibe coding.” Ms. Badani observed that AI tools are lowering the barrier to technical engagement, allowing non-engineers to become more technical, citing her son building an app. While not yet production-ready, this broadens the source of innovative ideas within organizations.

For infrastructure investment, Ms. Badani advised prioritizing “intelligence per watt” and efficiency. She predicted continued massive investments in data centers and the emergence of “AI everywhere,” including new device classes. The future infrastructure will be characterized by power-efficient, distributed computing.

Ms. Brown countered misinformation about AI-driven job losses. Upwork’s data indicates AI creates significantly more opportunities than it displaces, particularly for fractional and contingent workers. New roles, like “GTM engineer,” are rapidly emerging, highlighting an underestimation of AI’s capacity to generate new value and employment.

Both speakers offered an optimistic outlook. Ms. Badani predicted that AI agents would be integrated into organizational charts within 2-3 years, acting as job creators and force multipliers, driving significant GDP growth. Ms. Brown affirmed that the future of AI is human, with new opportunities for people outweighing job losses as the workforce reskills.