Geographic Hotspots for Agent-Company Funding: Where the GaaS Money Actually Lands

Table of Contents

• Why Geography Still Decides Who Gets Funded
• San Francisco: The Gravity Well
• The Second Tier Worth Watching
• New York: Vertical Agents Meet Enterprise Buyers
• London: Europe's Agent Capital
• Tel Aviv: Security and Reliability DNA
• Paris, Toronto, and the Research-Adjacent Cities
• The Emerging Challengers
• How to Read a Hotspot: The Four Ingredients
• What This Means If You're Raising
• Insights Most People Overlook
• References

Why Geography Still Decides Who Gets Funded

You'd think that in a market built on software that runs anywhere, location would have stopped mattering. It hasn't. If anything, agentic AI has re-concentrated capital around a few physical places, and the reason is worth sitting with for a second.

Agent companies are unusually dependent on three scarce inputs: researchers who understand how to make models behave reliably, early-adopter enterprise buyers willing to hand a workflow to software, and investors who can underwrite a business model that prices per task or per outcome rather than per seat. Those three things don't distribute evenly across the globe. They pool. And capital follows them into the pool.

There's also a less flattering reason. A lot of GaaS investing right now is pattern-matching. When a partner at a top firm sees three agent companies in the same zip code hitting the same revenue milestone, that becomes the template, and the next ten checks go to founders who look like those three. Proximity to the people writing the checks remains a quiet, powerful advantage, which is exactly why founders keep relocating to chase it. This dynamic shapes everything downstream, from Series A benchmarks for agent companies to the kind of due diligence late-stage investors run.

San Francisco: The Gravity Well

Let's not bury the lede. The San Francisco Bay Area is not one hotspot among several. It is the hotspot, and the gap between it and everywhere else is enormous.

The reasons compound on each other. The frontier model labs, the ones whose APIs every agent company builds on, are physically here. That means the engineers who left those labs to start agent companies are here too, and they bring relationships with the people still inside, which translates into early access to model features and, frankly, better intuition about where capability is heading. PitchBook and the National Venture Capital Association's data has shown for years that the Bay Area captures a wildly disproportionate slice of US venture dollars, and the agent wave has only sharpened that, as their Venture Monitor reports consistently document.

The density effect is real and hard to replicate. A founder in San Francisco can have coffee with three potential design partners, two ex-OpenAI engineers, and a Series A investor in a single afternoon, all within a few square miles. That's not a romantic story about serendipity, it's a measurable reduction in the time it takes to find your first ten customers and your next key hire. Andreessen Horowitz, which has been one of the loudest voices on agent infrastructure, has written repeatedly about why physical concentration accelerates company-building in exactly this category.

The downside, which everyone in the city will complain about without prompting, is cost. Senior AI engineers in the Bay command compensation that would fund an entire team elsewhere. That burn-rate pressure is a recurring theme across the cluster and connects directly to the burn-rate problem of running agents, when your talent and your inference bills are both denominated in San Francisco dollars, your default-alive math gets ugly fast.

The Second Tier Worth Watching

Below San Francisco sits a cluster of cities that each found an angle. None of them will out-raise the Bay in aggregate. But several are building real, defensible specializations, and for certain kinds of agent companies, they're arguably better places to be.

New York: Vertical Agents Meet Enterprise Buyers

New York's edge is its proximity to the buyers. Finance, legal, insurance, media, advertising, the industries most eager to deploy vertical agents have their decision-makers concentrated in Manhattan. A legal-agent startup or a financial-operations-agent startup can find design partners, pilot budgets, and eventually large contracts without ever leaving the island.

This matters more for GaaS than it did for traditional SaaS, because per-outcome pricing requires deep trust. You don't let an autonomous agent touch your invoices or your contracts on the strength of a demo. You let it in because you've spent months with the founders. New York's funding has skewed toward applied, vertical, revenue-first agent companies for exactly this reason, and that flavor of company tends to attract a different investor profile than the infrastructure plays out west.

London: Europe's Agent Capital

London is the clear European leader, and it's not especially close. It combines a deep pool of research talent, DeepMind's presence trained a generation of people who understand model behavior, with the continent's most mature venture market and an English-language advantage that lowers friction with US capital.

What's interesting about London is that its agent companies often go after global markets from day one, partly because the UK domestic market alone is too small to justify the valuations. That global orientation makes London startups attractive cross-border M&A targets, a thread that runs into cross-border GaaS M&A and regulatory review. The flip side is that European investors have historically priced AI risk more conservatively than their US counterparts, so London rounds tend to come at more grounded valuations, which some founders see as a feature, not a bug.

Tel Aviv: Security and Reliability DNA

Tel Aviv punches several weight classes above its size, and its specialization is unusually well-suited to this moment. The city's deep bench of cybersecurity and infrastructure talent, much of it forged in military intelligence units, maps almost perfectly onto the hardest unsolved problems in GaaS: agent security, sandboxing, permissioning, and reliability.

As enterprises wake up to the fact that an autonomous agent with credentials is also an autonomous attack surface, the companies solving agent security become essential infrastructure. Tel Aviv is producing a disproportionate share of them. Israeli founders also have a long track record of building for the US market while operating lean teams back home, a capital-efficiency story that resonates with investors burned by over-funded burn-heavy startups elsewhere.

Paris, Toronto, and the Research-Adjacent Cities

Paris rode the Mistral effect. The presence of a homegrown frontier lab gave the French ecosystem credibility, talent, and a magnet for capital it didn't have three years ago. French agent startups benefit from genuinely strong technical education and, increasingly, from European sovereignty narratives that channel both private and government-adjacent money toward locally controlled AI.

Toronto and the broader Canada corridor lean on their academic pedigree, the Vector Institute and the deep-learning research lineage tied to it, plus a cost structure meaningfully below San Francisco. The pattern in these research-adjacent cities is consistent: strong technical founding teams, somewhat thinner local growth capital, and a tendency to raise later-stage money from US funds once they've proven the technology. They're great places to start an agent company and, historically, harder places to scale one without going abroad for capital.

The Emerging Challengers

A few geographies are early but worth keeping on the radar. Bangalore and the broader Indian ecosystem combine enormous engineering talent with a domestic enterprise market that's increasingly willing to adopt agents, and the cost arbitrage is severe. Singapore is positioning itself as the agent hub for Southeast Asia, with government money actively courting the category. Berlin retains a steady stream of technical founders, though it sits in London's shadow for European capital.

The honest caveat: "emerging" often stays emerging. Plenty of cities have been the next big thing for AI funding without ever graduating. The signal to watch isn't the first splashy round, it's whether a second and third company in the same city can raise on the back of the first one's success. Ecosystems are built on local proof points, and a single unicorn doesn't make a hotspot. McKinsey's ongoing work on where AI value and investment are concentrating is a useful reality check against the louder regional boosterism.

How to Read a Hotspot: The Four Ingredients

Strip away the local marketing and every genuine agent-funding hotspot has the same four ingredients. If you're evaluating a city, as a founder deciding where to base, or an investor deciding where to spend time, check for all four.

First, research talent with model-behavior expertise, usually downstream of a nearby lab or a strong university program. Building agents that don't go off the rails is a research problem, not just an engineering one.

Second, early-adopter enterprise buyers within reach. GaaS lives or dies on landing real workflows, and the cities that win tend to sit next to the industries most willing to experiment.

Third, investors who understand outcome-based pricing. A surprising number of otherwise sophisticated firms still try to value agent companies on SaaS comps, which is why understanding how VCs underwrite GaaS differently from SaaS is itself a geographic variable, some cities have it, most don't.

Fourth, a cost structure the business model can survive. Agents are expensive to run, margins compress when model prices move, and a city where your all-in cost per employee is double the competition's is a structural headwind no amount of local pride overcomes.

When all four line up, you get a hotspot. When one is missing, you get a place that produces good founders who eventually leave.

What This Means If You're Raising

If you're a founder reading this as a tactical question, where should I be?, the answer is less binary than it used to be. The genuinely useful framing is to separate where you build from where you raise.

Plenty of strong agent companies now build in lower-cost research cities and raise from San Francisco and New York money, traveling for the rounds and the key relationships. That hybrid is increasingly normal and, in a capital-efficiency-conscious market, often the smart play. What hasn't changed is that being a complete unknown to the firms writing the largest agent checks is a real disadvantage. The relationships still cluster geographically even when the engineering doesn't.

The other shift worth internalizing: investors are getting more discerning about which regional stories are real. Early in the cycle, simply being an "AI agent company in [hot city]" was enough to get meetings. That window is closing. The cities that keep their hotspot status will be the ones where the four ingredients above genuinely compound, not the ones that issued the most press releases. Watch the second and third funded companies in any city, they tell you far more than the first.

Insights Most People Overlook

The hotspot map for funding doesn't match the map for building. Most coverage conflates where agent companies are headquartered with where they raise money. They're diverging. A growing share of teams build in Toronto, Tel Aviv, Bangalore, or Paris and raise primarily from Bay Area and New York funds. If you only track headquarters data, you'll badly misread where capital actually originates, and you'll underrate cities that are talent exporters but capital importers.

Specialization beats size for everyone except San Francisco. The second-tier cities that are thriving didn't try to be mini-San Franciscos. They went deep on one thing, Tel Aviv on security, New York on regulated verticals, London on global-from-day-one. The cities trying to be generalist AI hubs are the ones losing momentum. For a founder, this means the relevant question isn't "is this a good AI city" but "is this the right city for my specific kind of agent company."

Proximity to model labs is a depreciating asset. Right now, being near a frontier lab confers a real edge, early feature access, recruiting pipelines, better roadmap intuition. But as model capabilities commoditize and the differentiation moves up to the application and reliability layer, that geographic premium will erode. Cities betting their entire identity on lab-adjacency may find the advantage thinner in a few years than it looks today. The durable edge is buyer-adjacency, not lab-adjacency.

Conservative European valuations may age better than they look. Founders often gripe that London and Paris price rounds below San Francisco. But in a market openly debating a GaaS valuation bubble, raising at a defensible valuation is insurance against the down-round risk that's stalking over-funded startups. The "lower" valuation hotspots may simply have fewer companies facing painful resets when the market recalibrates.

The real geographic risk is single-ecosystem fragility. Cities whose entire agent scene rests on one breakout company or one lab are fragile. If that anchor stumbles, talent disperses, the narrative dies, the whole local funding story can evaporate. Diversified ecosystems with multiple independent success stories are far more durable, which is another reason San Francisco's depth is so hard to challenge: no single failure can unmake it.

References

• National Venture Capital Association, Research and Venture Monitor Reports
• Andreessen Horowitz, The Economic Case for Generative AI and Foundation Models
• McKinsey & Company, The State of AI