The conventional expectation about AI and real estate brokerage was that it would flatten the playing field. Tools that were once available only to firms with substantial technology budgets would become accessible to anyone with a subscription, and the information advantage that large brokerages had accumulated over decades would erode. Some of that has happened. But the more significant effect has been closer to the opposite, and the actual picture emerging in 2026 is stranger and more interesting than either the consolidation thesis or the democratization thesis predicted on its own.

The scale argument is easy to see in the spending. CBRE reported nearly $1.7 billion on computer hardware and software in its 2025 annual report, up $300 million from the year before. On the firm’s fourth-quarter earnings call, CEO Robert Sulentic said AI integration is projected to drive a 25% reduction in research costs, and described what the company expects by the end of 2026. “That is being enabled by AI and that’s one of the areas we’re most encouraged today,” Sulentic said. “It’s going to save us money in terms of accumulating the data, buying the data, and it’s going to make our brokers more efficient in terms of using the data. We’re also using that same set of tools to meaningfully cut the cost of our research efforts.” CBRE has already achieved a 25% reduction in manual lease processing time and deployed AI across more than a billion square feet of managed property.

JLL has been making a similar argument from the margin side. CEO Christian Ulbrich told investors that AI has helped drive strong margin performance over the past two years as the company becomes more productive across its business lines. The firm has invested roughly $450 million across 55 proptech startups through JLL Spark since 2018, including an early stake in Dealpath that is now used across the industry by firms such as CBRE and Cushman & Wakefield. Cushman & Wakefield, working to close the gap, has partnered with Microsoft on AI solutions rather than building comparable infrastructure independently.

Meanwhile, the smallest firms are falling behind on basic adoption. Delta Media Group’s 2026 analysis found that brokerages with more than 100 agents and those with 11 to 50 agents reported 100% agent AI usage, while firms with 10 or fewer agents reported just 81.8%. Independent shops with single-digit agent counts remain, in the analysis’s phrasing, meaningfully behind, and they posted the highest non-adoption rate in the industry at 9.1%. Industry-wide resistance has collapsed, with the share of brokerages using no AI at all falling from 24.8% in 2024 to 3.9% in 2026, but the smallest firms are the ones left in that shrinking group.

And yet the more surprising development is happening at the other end of the same spectrum, where AI is creating a channel that scale does not automatically win. As investors increasingly ask generative AI tools which brokers to call, a new form of competition has emerged around what those tools recommend. ViewEO found that in Brooklyn, a small boutique brokerage, Terra CRG, was outperforming the top five largest firms in AI recommendations for that market. The reason had nothing to do with technology budget. Terra CRG specialized in Brooklyn, transacted exclusively there, and consistently published its deals and market commentary. The models, asked who knows Brooklyn commercial real estate, found a firm whose entire public footprint answered that question.

That mechanism is genuinely new. “I would say the biggest challenge for a lot of brokers who are not 30 years in the business is just brand awareness, awareness of what you do, what differentiates you,” said Taylor Avakian, a first vice president at Lyon Stahl Investment Real Estate. AI recommendation systems reward clarity of specialization in a way that traditional brand-building did not. A national firm with a presence in every market and every property type produces a diffuse signal. A firm that does one thing in one place produces a sharp one. Being the largest brokerage in the country does not make you the most obvious answer to a narrow question.

The independent segment is also getting purpose-built infrastructure for the first time. Tim Rodland, founder of Rodland Real Estate, framed the historical constraint bluntly. “I think that most of the challenges that many independents have is they don’t have access to technology. Independents had two opportunities back in the day, or up to this point: stay small or join another larger system. Those were their two options. We’re trying to scale independence by offering technology that was not available before.” His firm’s RoRo platform delivers real-time market intelligence through a conversational interface, aimed at giving boutique firms analytical capability without requiring them to affiliate with a franchise to get it.

Whether that access is enough to matter is the open question. The tools are real and the recommendation channel is real, but neither replicates the proprietary transaction data that gives the largest firms their compounding advantage. What the small-firm playbook does offer is a different basis for competition entirely. Depth in a defined market, a consistent public record of activity in it, and analytical tools that are good enough rather than best in class may produce more AI visibility than breadth backed by a nine-figure technology budget.

The industry that results is likely to be more barbell-shaped than it was. The largest firms are pulling further ahead on efficiency, margin, and data assets, and their acquisitions are increasingly justified by the datasets they absorb. Firms with genuine specialization and disciplined public presence have a new channel that does not require matching that spending. The position that looks hardest to hold is the middle, where a firm is too large to be legible as a specialist and too small to fund the data flywheel. That has always been an uncomfortable place to sit in brokerage. AI is making it more so.

The post AI Is Making Big Brokerages Bigger and Small Ones More Visible appeared first on Propmodo.

​