Dream, an AI and cyber defense company built for governments and critical infrastructure operators, announced a $260 million funding round on June 18 that values the company at $3 billion. The round was co-led by Bicycle Capital and Group 11, with participation from Antler, Bain Capital Ventures, Tru Arrow Partners and other investors.
The company says the new money will support expansion across Europe, the Middle East, Asia and the Americas. Dream also says it has secured nearly $300 million in total contract value since beginning commercial operations in late 2024, a striking claim for a company founded in 2023.
The headline number matters, but the sharper signal is what investors are funding. Dream is not selling a general enterprise chatbot, a consumer assistant or another cloud security dashboard. Its pitch is that governments need AI systems and cyber defense platforms they can operate under their own authority, in controlled environments, with national data kept inside sovereign boundaries.
What Dream Is Selling
Dream describes three platforms. Sphere is its national cyber defense system, combining cyber intelligence, exposure management, attack-path analysis, digital twins and AI-powered detection and response. Hero is an autonomous AI security researcher designed to find vulnerabilities, reason through attack paths and stress-test defenses before hostile actors exploit them. Atlas is the sovereign AI platform, intended to connect fragmented government data, structure it into usable knowledge and support mission-specific agents and models inside secure government-controlled environments.
That product split explains why the company sits between several markets that are usually covered separately: cyber command systems, critical-infrastructure defense, government data platforms and sovereign AI. Dream is arguing that those markets are converging. A country that cannot protect its data, reason over it or run models against it without depending on foreign vendors has a technology dependency problem as much as a cybersecurity problem.
The company was founded by Shalev Hulio, former Austrian Chancellor Sebastian Kurz and Gil Dolev. Dream says it now employs about 350 people across Tel Aviv, Abu Dhabi and Vienna, with customers in Europe, the Middle East and Southeast Asia. Hulio’s previous role as co-founder and chief executive of NSO Group, the Israeli company known for the Pegasus spyware controversy, gives Dream both deep offensive-security pedigree and a scrutiny problem that ordinary cyber startups do not face.
Why Governments Are The Target Customer
Most enterprises can choose between cloud security vendors, managed detection providers and AI platforms from major U.S., European or Chinese technology companies. Governments have a harder calculation. Defense agencies, intelligence services, health ministries, energy operators and national cyber authorities often hold data that cannot be moved freely across borders or exposed to opaque model-training, support or telemetry pipelines.
That is where the phrase “sovereign AI” has moved from policy language into procurement. It can mean many things, including local data residency, national cloud controls, models hosted inside a country, locally governed identity systems, export-control resilience or public-sector ability to inspect and direct the technology stack. Dream’s version is more security-centered: keep the data, models and cyber operations under national control, then use AI to detect threats, find weaknesses and turn disconnected government information into operational intelligence.
The timing helps. Governments are already worried about AI model access, cloud dependence, cyberattacks on energy and telecom networks, and the possibility that future AI capabilities may be restricted by export controls or vendor policy. Critical-infrastructure operators also face a more immediate problem: they need faster vulnerability discovery and threat detection without sending sensitive operational data into tools they cannot fully govern.
The Hard Part Is Trust
Dream’s funding round shows investor confidence in a category, not proof that the category is solved. Government-owned AI and cyber platforms have to clear a higher bar than conventional software. Buyers will need to know where models run, how data is isolated, who can access logs and telemetry, whether automated security research can be constrained, and how findings are reviewed before they turn into operational changes.
Those questions become sharper when a system claims to reason like an adversary or continuously stress-test defenses. In a national cyber environment, a false positive can waste scarce analyst time, but an overconfident automated action can be worse. Useful AI security systems need audit trails, permission boundaries, human review points, red-team evidence and clear rules for what the agent can touch.
There is also a geopolitical sales challenge. A platform marketed as sovereign infrastructure has to persuade governments that dependence on one vendor is not simply replacing one form of dependence with another. That means architecture, contractual control, local operations, support access and exit paths matter almost as much as model performance.
For now, Dream’s $260 million round is one of the clearest signs that AI cybersecurity for governments is becoming an infrastructure market. The company is betting that national cyber defense will not be a normal SaaS category. It will be sold as a controlled stack for countries that want AI capability, cyber resilience and data sovereignty packaged together.
Sources: Dream announcement via AIwire, SecurityWeek, Dream company profile, Business Insider.
