Build or buy? Scaling your enterprise gen AI pipeline in 2025

Scaling adoption of generative tools has always been a challenge of balancing ambition with practicality, and in 2025, the stakes are higher than ever. Enterprises racing to adopt large language models (LLMs) are encountering a new reality: Scaling isn’t just about deploying bigger models or investing in cutting-edge tools — it’s about integrating AI in ways that transform operations, empower teams and optimize costs. Success hinges on more than technology; it requires a cultural and operational shift that aligns AI capabilities with business goals.

The scaling imperative: Why 2025 is different

As generative AI evolves from experimentation to enterprise-scale deployments, businesses are facing an inflection point. The excitement of early adoption has given way to the practical challenges of maintaining efficiency, managing costs and ensuring relevance in competitive markets. Scaling AI in 2025 is about answering hard questions: How can businesses make generative tools impactful across departments? What infrastructure will support AI growth without bottlenecking resources? And perhaps most importantly, how do teams adapt to AI-driven workflows?

Success hinges on three critical principles: identifying clear, high-value use cases; maintaining technological flexibility; and fostering a workforce equipped to adapt. Enterprises that succeed don’t just adopt gen AI — they craft strategies that align the technology with business needs, continually reevaluating costs, performance and the cultural shifts required for sustained impact. This approach isn’t just about deploying cutting-edge tools; it’s about building operational resilience and scalability in an environment where technology and markets evolve at breakneck speed.

Companies like Wayfair and Expedia embody these lessons, showcasing how hybrid approaches to LLM adoption can transform operations. By blending external platforms with bespoke solutions, these businesses illustrate the power of balancing agility with precision, setting a model for others.

Combining customization with flexibility

The decision to build or buy gen AI tools is often portrayed as binary, but Wayfair and Expedia illustrate the advantages of a nuanced strategy. Fiona Tan, Wayfair’s CTO, underscores the value of balancing flexibility with specificity. Wayfair uses Google’s Vertex AI for general applications while developing proprietary tools for niche requirements. Tan shared the company’s iterative approach, sharing how smaller, cost-effective models often outperform larger, more expensive options in tagging product attributes like fabric and furniture colors.

Similarly, Expedia employs a multi-vendor LLM proxy layer that allows seamless integration of various models. Rajesh Naidu, Expedia’s senior vice president, describes their strategy as a way to remain agile while optimizing costs. “We are always opportunistic, looking at best-of-breed [models] where it makes sense, but we are also willing to build for our own domain,” Naidu explains. This flexibility ensures the team can adapt to evolving business needs without being locked into a single vendor.

Such hybrid approaches recall the enterprise resource planning (ERP) evolution of the 1990s, when enterprises had to decide between adopting rigid, out-of-the-box solutions and heavily customizing systems to fit their workflows. Then, as now, the companies that succeeded recognized the value of blending external tools with tailored developments to address specific operational challenges.

Operational efficiency for core business functions

Both Wayfair and Expedia demonstrate that the real power of LLMs lies in targeted applications that deliver measurable impact. Wayfair uses generative AI to enrich its product catalog, enhancing metadata with autonomous accuracy. This not only streamlines workflows but improves search and customer recommendations. Tan highlights another transformative application: leveraging LLMs to analyze outdated database structures. With original system designers no longer available, gen AI enables Wayfair to mitigate technical debt and uncover new efficiencies in legacy systems.

Expedia has found success integrating gen AI across customer service and developer workflows. Naidu shares that a custom gen AI tool designed for call summarization ensures that “90% of travelers can get to an agent within 30 seconds,” contributing towards a significant improvement in customer satisfaction. Additionally, GitHub Copilot has been deployed enterprise-wide, accelerating code generation and debugging. These operational gains underscore the importance of aligning gen AI capabilities with clear, high-value business use cases.

The role of hardware in gen AI

The hardware considerations of scaling LLMs are often overlooked, but they play a crucial role in long-term sustainability. Both Wayfair and Expedia currently rely on cloud infrastructure to manage their gen AI workloads. Tan notes that Wayfair continues to assess the scalability of cloud providers like Google, while keeping an eye on the potential need for localized infrastructure to handle real-time applications more efficiently.

Expedia’s approach also emphasizes flexibility. Hosted primarily on AWS, the company employs a proxy layer to dynamically route tasks to the most appropriate compute environment. This system balances performance with cost efficiency, ensuring that inference costs don’t spiral out of control. Naidu highlights the importance of this adaptability as enterprise gen AI applications grow more complex and demand higher processing power.

This focus on infrastructure reflects broader trends in enterprise computing, reminiscent of the shift from monolithic data centers to microservices architectures. As companies like Wayfair and Expedia scale their LLM capabilities, they showcase the importance of balancing cloud scalability with emerging options like edge computing and custom chips.

Training, governance and change management

Deploying LLMs isn’t just a technological challenge — it’s a cultural one. Both Wayfair and Expedia emphasize the importance of fostering organizational readiness to adopt and integrate gen AI tools. At Wayfair, comprehensive training ensures employees across departments can adapt to new workflows, especially in areas like customer service, where AI-generated responses require human oversight to match the company’s voice and tone.

Expedia has taken governance a step further by establishing a Responsible AI Council to oversee all major gen AI-related decisions. This council ensures that deployments align with ethical guidelines and business objectives, fostering trust across the organization. Naidu underscores the significance of rethinking metrics to measure gen AI’s effectiveness. Traditional KPIs often fall short, prompting Expedia to adopt precision and recall metrics that better align with business goals.

These cultural adaptations are critical to gen AI’s long-term success in enterprise settings. Technology alone cannot drive transformation; transformation requires a workforce equipped to leverage gen AI’s capabilities and a governance structure that ensures responsible implementation.

Lessons for scaling success

The experiences of Wayfair and Expedia offer valuable lessons for any organization looking to scale LLMs effectively. Both companies demonstrate that success hinges on identifying clear business use cases, maintaining flexibility in technology choices, and fostering a culture of adaptation. Their hybrid approaches provide a model for balancing innovation with efficiency, ensuring that gen AI investments deliver tangible results.

What makes scaling AI in 2025 an unprecedented challenge is the pace of technological and cultural change. The hybrid strategies, flexible infrastructures and strong data cultures that define successful AI deployments today will lay the groundwork for the next wave of innovation. Enterprises that build these foundations now won’t just scale AI; they’ll scale resilience, adaptability, and competitive advantage.

Looking ahead, the challenges of inference costs, real-time capabilities and evolving infrastructure needs will continue to shape the enterprise gen AI landscape. As Naidu aptly puts it, “Gen AI and LLMs are going to be a long-term investment for us and it has differentiated us in the travel space. We have to be mindful that this will require some conscious investment prioritization and understanding of use cases.”