Oxylabs experts forecast AI bubble risks & browser wars

Oxylabs has set out a series of divergent predictions for artificial intelligence in 2026, highlighting the risk of an investment bubble, disputes over European data rules, and a new wave of AI-driven web browsers.

The Vilnius-based web intelligence company gathered views from internal leaders and external AI specialists. The group included its chief executive, governance chief, and members of its AI and machine learning board. A former NASA data scientist also contributed.

The experts expect strong investment in AI to continue in the near term. They also warn about unintended outcomes from regulation and shifts in how users access and search the internet.

Bubble concerns

Oxylabs chief executive Julius Černiauskas said talk of an AI bubble will persist. He does not expect it to burst in the coming year.

"It is likely that in 2025, narratives around the AI bubble will continue to surface - but it probably won't burst next year. The advancements we are seeing companies make are enough to secure an investment influx and protect the industry. Clearly, many are seeing gains with this technology, enough to fend off speculation.

"While AI may not have produced the expected ROI just yet, the money will likely continue to come in, even with the incremental improvements in AI. In the next few years, we will see fears be replaced by impact-heavy case studies which are able to speak to the power of AI when applied thoughtfully," said Černiauskas.

The group sees a shift in how enterprises use AI inside their organisations. They describe a move from narrow assistance towards systems that run whole workflows.

Enterprise workflows

Ali Chaudhry is a member of the Oxylabs AI/ML board, a researcher, and founder of DataLumio. He said the coming year will mark a turning point for adoption across corporate functions.

"2026 will be the year that AI really starts to make an impact at enterprises in every part of the organisation. We will see AI agents transition from being siloed helpers to being able to complete a whole chain of commands, such as 'send this report to all involved parties and file away'," said Chaudhry.

He linked this shift with growing regulatory pressure in major markets.

"But this isn't the only place that workers will feel the impact of AI, as a greater need to comply with growing legislation, think EU AI Act, will cause leaders to have to pivot and build this into their existing governance structure. This means we will also see the ROI become more measurable, and speculation around the technology decrease.

"Finally, we will start to see AGI transition to become more user-friendly by having safety measures put in place. There will be dedicated 'Safety & Alignment' budgets, third-party audits, and pre-deployment risk assessments for frontier releases as we work out what it means to operationalise such a ground-breaking technology," said Chaudhry.

Data and regulation

Oxylabs' chief governance and strategy officer, Denas Grybauskas, raised concerns about the impact of European data rules on AI development. His comments focus on restrictions affecting public data.

"Next year, if we continue to see companies in Europe backed into a corner and unable to collect enough data due to restrictions on public data, they might push forward with biased AI models built on small data sets," said Grybauskas.

He said tougher rules could also change how the largest technology firms respond to European policymakers.

"Some overly burdensome legislation might even drive Big Tech companies, such as those in the USA and China, to lobby for exemptions. Already across Europe, discussions favoring deregulation are starting to take shape. The conversation will intensify next year, and the first concrete proposals are likely to emerge," said Grybauskas.

'New order' fears

Another expert highlighted broader social and economic risks. Adi Andrei is a member of the Oxylabs AI/ML board and director at Technosophics. He previously worked as a senior data scientist at NASA.

"In the next three years, we should expect incremental improvements in AI model performance: slightly better coherence, faster inference, and improved multimodal integration. Meanwhile, the hype machine will intensify. Venture capital, media narratives, and corporate PR will continue to inflate expectations - creating a speculative bubble that may eventually burst with catastrophic consequences for markets and jobs. But this could create the perfect opportunity to enforce the new system," said Andrei.

He linked this risk with changes in global finance and digital control systems.

"We are witnessing the quiet dismantling of the old global financial order - and its replacement with a new, digitally enforced regime: programmable money, algorithmic censorship, predictive policing, and behavioral nudging - all under the banner of 'efficiency' and 'security'," said Andrei.

Browser competition

The experts also expect AI integration in browsers to reshape the consumer internet. This includes new competitors for the dominant Chrome browser.

Rytis Ulys, head of data and AI at Oxylabs, pointed to a new generation of AI-native browsers.

"I'm really looking forward to seeing how AI-native browsers affect what services people choose to use in 2026. The launch of AI-native browsers from OpenAI's Atlas to Perplexity's Comet marks the first serious challenge to Chrome's 65% market share in over a decade. These aren't just browsers with AI features bolted on - they're fundamentally reimagined interfaces where the browser acts as an autonomous agent that completes tasks, manages workflows, and serves as a personal research assistant," said Ulys.

He said privacy concerns will shape the way these tools handle data.

"To win the browser wars on privacy, AI models will increasingly run on-device. These 'Edge AI' models will read, summarise, and 'collect' web content locally into personal knowledge graphs. This trend will decentralise data collection, making it privacy-centric but also creating a new, untraceable form of data extraction," said Ulys.

Parsing and web data

Oxylabs' chief operating officer, Juras Juršėnas, focused on large language model use in web data parsing. He said falling costs and technical changes will alter workflows for developers and data teams.

"Over the next 12 months, the use of LLMs for parsing will grow. For the last few years, data parsing has been one of the most impactful use cases for AI in public data collection. However, it was still limited by price (for LLM tokens) and prompt size limitations. Developers and data teams used to always need to clean the HTML to reduce its size before giving it to the LLM to parse, which required additional resources. Now you might only need to do this in specific cases. The number of options in the market for tools that can do it for you is booming. Thus, it is reasonable to expect an increase in LLM usage for parsing," said Juršėnas.

The experts expect these trends in investment, regulation, browsing, and data use to shape how AI develops through 2026 and the years that follow.