en.Wedoany.com Reported - During the AI Engineer World's Fair held in San Francisco, Oxylabs CEO Vytautas Savickas pointed out that the biggest shift in the industry is not occurring within the models themselves, but at the infrastructure level. Savickas stated that the focus of AI development over the past three years has been on better models, while the next phase will revolve around the systems and real-time information surrounding these models.
AI systems no longer rely solely on knowledge acquired during training; they increasingly require fresh information, real-time search, and reliable access to the ever-changing web. Savickas believes that knowledge is not static, and models not connected to new information have a far lower understanding of the current world than most people realize.
Every breakthrough in AI has transformed the underlying infrastructure. From the need for vast and diverse public data to train foundation models, to the requirement of Retrieval-Augmented Generation (RAG) to understand changes that occurred five minutes ago, and to the agent era where systems must perform searches, comparisons, and verifications on behalf of users, each wave imposes different demands on infrastructure. Public web data is evolving from training material into a component of AI runtime.
Oxylabs began building large-scale public web data infrastructure long before large models became mainstream. The company currently serves over 15,000 clients globally, holds more than 160 patents, and operates one of the world's largest public web data infrastructures.
Savickas believes that model quality is no longer a differentiating factor in many practical applications; how reliably AI systems connect to the external world has become more critical. He noted that the reason models hallucinate is often due to reasoning with outdated or incomplete information, rather than a lack of intelligence. He concluded that the companies winning in AI are not necessarily those building the largest models, but those building the systems users trust the most.
The rise of AI agents is changing how engineers think about infrastructure. Savickas stated that factors such as latency, reliability, and browser automation, while not often making headlines, are key to determining whether AI is truly effective. The open web, as humanity's largest shared knowledge resource, should remain accessible to support technological innovation.
Over nearly a decade, Oxylabs has witnessed multiple waves of technology, from travel aggregation and digital marketing to e-commerce and cybersecurity. Savickas believes that the first generation of AI proved machines can reason; the next challenge is to make these systems operate reliably in the real world. The future of AI will be driven by both better models and the superior infrastructure surrounding them.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
