The surge in generative AI rapidly spawned countless new ventures. However, as the initial frenzy subsides, two formerly trending business paradigms are appearing increasingly as warnings: LLM wrappers and AI aggregators.
Darren Mowry, who oversees Google’s worldwide entrepreneurial division spanning Cloud, DeepMind, and Alphabet, suggests that ventures employing these strategies are signaling distress.
LLM wrappers are fundamentally new companies that encase pre-existing expansive language models, such as Claude, GPT, or Gemini, by adding a product or user-facing layer to address a particular issue. For instance, a venture leveraging AI to assist learners in their academic pursuits would be one such example.
“If you’re solely relying on the underlying model to perform every function and effectively rebranding that model, the sector no longer tolerates such an approach,” Mowry expressed during this week’s installment of Equity.
Mowry asserts that applying “scant original content layered over Gemini or GPT-5” indicates a lack of unique distinction.
“You need to establish profound, broad competitive advantages that are either distinctive across various applications or highly tailored to a niche sector,” he commented, “for a new venture to ‘advance and expand’.” Illustrative deep-moat LLM wrapper firms encompass Cursor, a coding helper driven by GPT, or Harvey AI, an AI-driven legal support tool.
Techcrunch event
Boston, MA
|
June 9, 2026
To put it differently, new ventures should no longer anticipate merely overlaying a user interface onto a GPT model and gaining significant adoption for their offering, as might have been feasible, for instance, in mid-2024 upon the introduction of OpenAI’s ChatGPT marketplace. The current imperative involves forging enduring intrinsic worth for the product.
AI consolidators represent a specific category of wrappers — they are fledgling companies that combine numerous LLMs within a unified interface or API stratum for channeling requests across diverse models and granting end-users entry to several models. Such entities usually offer a coordination stratum that encompasses oversight, regulatory control, or evaluation utilities. Consider: the AI search venture Perplexity or the developer ecosystem OpenRouter, which offers entry to numerous AI models through a singular API.
Although several of these frameworks have achieved some momentum, Mowry’s counsel is unequivocal for emerging new ventures: “Avoid the aggregation sector.”
Broadly speaking, aggregators are not experiencing significant expansion or advancement currently because, he contends, consumers desire “integrated proprietary content” to guarantee appropriate model redirection precisely when required, in alignment with their demands — rather than due to underlying computational or availability limitations.
Mowry possesses extensive experience in the cloud domain, spanning decades, having gained initial experience at AWS and Microsoft prior to establishing his presence at Google Cloud, and he has observed the unfolding of these dynamics. He drew parallels between the present circumstances and the nascent period of cloud computing during the late 2000s and early 2010s when Amazon’s cloud operations began to flourish.
During that era, a multitude of nascent companies emerged to re-offer AWS foundational services, positioning themselves as more accessible gateways that furnished utilities, unified invoicing, and assistance. However, once Amazon developed its proprietary corporate solutions and clients became adept at overseeing cloud offerings firsthand, the majority of those fledgling companies were displaced. Only those that introduced genuine value-added services persisted, such as cybersecurity, system transfers, or DevOps advisory.
Present-day AI consolidators encounter comparable profit margin strain because core model developers are venturing into enterprise-level capabilities independently, thereby potentially marginalizing intermediaries.
As for Mowry, he remains optimistic about intuitive coding and development ecosystems, which experienced an unprecedented year in 2025, with new ventures such as Replit, Lovable, and Cursor (all clients of Google Cloud, according to Mowry) garnering substantial funding and user engagement.
Mowry further anticipates robust expansion in direct-to-client technology, specifically within firms that equip end-users with these potent AI instruments. He highlighted the potential for cinema and television scholars to leverage Google’s AI-powered video creation tool Veo to actualize narratives.
Beyond the realm of AI, Mowry also believes that biotechnology and climate technology are experiencing a period of prominence — owing to both the infusion of venture capital into these two sectors and the “vast quantities of information” new ventures can utilize to generate tangible worth “through methods previously unattainable.”
{content}
Source: {feed_title}