A month after the Hugging Face breach, new details reveal AI agents coordinated, rebuilt deleted infrastructure and escalated access in hours. The fallout is now reaching Congress, regulators and frontier labs, raising urgent questions about AI security and control.
A Zhuhai operator used DeepSeek, a Hermes agent and a godmode jailbreak to hunt 460-plus targets. Open-weight models made the offensive kit a shopping list.
A Zhuhai operator used DeepSeek, a Hermes agent and a bundled godmode jailbreak to hunt more than 460 targets, then stole data from three organisations. Open-weight models this week made that offensive kit a public shopping list for any lone operator who can still point one at his scanner tonight.
Australia's Agentic Awakening: Startups Step into the AI Endgame
Australia is entering the age of agentic intelligence as startups like Firmus Technologies and Sharon AI build sovereign compute, renewable powered data infrastructure and AI platforms. Infrastructure is accelerating while enterprise adoption remains slow, creating a widening national gap.
As the global AI race hits its final stretch in late 2025, Australia's tech scene is finding its own pace – not by chasing Silicon Valley, but by building what comes after it. A wave of startups is rising to meet the next frontier: agentic intelligence. Their mission isn't to replicate the world's biggest models but to orchestrate the systems that will drive them – autonomous, connected, and sovereign by design.
The mood within Australian AI circles has changed. Where 2023 and 2024 were about building copilots and model wrappers, 2025 has become the year of orchestration. The real prize is no longer a chatbot, but an entire network of goal-seeking agents – intelligent systems that negotiate, optimise and create with minimal human oversight. For a country that prides itself on pragmatism, this shift resonates. The ambition is straightforward: own the infrastructure, secure the sovereignty, and scale the intelligence.
That's where companies like Firmus Technologies and Sharon AI have taken the lead. Firmus has become the name most often whispered when investors talk about sovereign compute. Now valued above A$1.85 billion, its Project Southgate aims to knit together a chain of renewable-powered data centres across the nation, anchored by NVIDIA's incoming GB300 GPUs.
The first phase, in Tasmania, is already running 18,500 units and ties into 5.1GW of renewable generation scheduled through 2028. It's a playbook rooted in sovereignty and scale. By coupling green energy with national compute, Firmus has positioned itself as the architecture beneath the new AI economy.
As CEO Oliver Curtis puts it, Southgate isn't another hyperscaler – it's the "infrastructure blueprint for scalable, sovereign intelligence," designed to keep the country and its partners on the frontline of the global data economy.
If Firmus builds the foundation, Sharon AI is the platform that rides atop it. The company is preparing a pre-IPO raise ahead of a planned 2026 listing, backed by a new 50MW partnership with NEXTDC – enough to host around 20,000 next-generation GPUs across Tier IV facilities.
Co Founders of Sharon AI - Nick Hughes-Jones and Andrew Leece. Source: Sharon AI
Its pitch is as pragmatic as it is ambitious: an "agentic cloud" built for flexible GPU leasing, rapid orchestration and burst capacity for enterprises, governments and researchers. Underpinning it all are NVIDIA DGX-Ready credentials, ensuring new hardware lands ready-to-train without delays or retrofits. CEO Wolf Schubert sums it up succinctly:
"Autonomous systems need autonomous infrastructure. We're building exactly that."
Both Firmus and Sharon AI are currently in advanced talks with international investment firms to lead what could become the largest AI mega-scale IPO the ASX has ever seen. Financial media has been keeping a close eye on the developments, with analysts speculating that the combined market debut in 2026 could redefine the scale of Australian tech listings. Yet the infrastructure boom tells only half the story. While compute capacity heats up, the reality on the ground remains decidedly cooler.
The Reserve Bank of Australia's latest survey results, released today, paint a more measured picture of where enterprise adoption actually stands. Beyond the AI chatbot or the now-ubiquitous ChatGPT wrapper, genuine integration of intelligent systems into Australian business operations remains lacklustre. The RBA's April 2025 Financial Stability Review flags a widening gap between infrastructure readiness and real-world deployment, with most firms still treating AI as experimental rather than operational. The federal government's messaging around a "one voice approach" to AI adoption in society has yet to translate into the coordinated, economy-wide transformation that infrastructure builders are anticipating. In practice, it's slow as you go.
The RBA's core concerns centre on three interlocking risks that could complicate the path ahead. First, the concentration of AI capabilities within a handful of large technology firms creates systemic dependencies that could amplify shocks across the financial system. Second, operational resilience remains fragile as organisations rush to adopt AI tools without fully understanding their failure modes or embedding proper governance frameworks. Third, cyber vulnerabilities multiply as AI systems become embedded in critical financial infrastructure, creating new attack surfaces that regulators are still learning to monitor. Together, these concerns underscore a fundamental tension: the infrastructure is being built at breakneck speed, but the institutional capacity to deploy it safely and effectively is lagging well behind.
Yet private capital continues to heat up. Sharon's convertible note raise has drawn institutional attention from funds positioning for the next compute cycle, while its partnership web – stretching from NEXTDC and NVIDIA to advisers like Roth CH – signals confidence that demand will eventually catch up to supply. The new wave of Australian AI firms is merging industrial hardware, smart energy, and software agency into one machine. The result: a national compute layer capable of both accelerating industry and exporting intelligence abroad. From climate modelling to financial automation, from health research to defence applications, the stakes are now measured in teraflops and trust. Australia's next leap won't come from building larger models, but from enabling intelligent systems that act, adapt and learn across sovereign infrastructure. In other words, from agentic scale.
The AI race may be ending, but the age of agency has just begun – and this time, Australia is not catching up. It's showing up. The question now is whether federal capital partnerships can bridge the gap between infrastructure ambition and enterprise reality, bringing forward an entire AI ecosystem rather than isolated pockets of capability. The foundations are being laid. Whether the building follows remains the A$1.85 billion question.
Get the stories that matter to you. Subscribe to Cyber News Centre and update your preferences to follow our Daily 4min Cyber Update, Innovative AI Startups, The AI Diplomat series, or the main Cyber News Centre newsletter — featuring in-depth analysis on major cyber incidents, tech breakthroughs, global policy, and AI developments.
Sign up for Cyber News Centre
Where cybersecurity meets innovation, the CNC team delivers AI and tech breakthroughs for our digital future. We analyze incidents, data, and insights to keep you informed, secure, and ahead.
NVIDIA is asking Wall Street to underwrite AI compute as productive infrastructure. Its new financing platforms aim to mobilise more than US$500 billion, but the announced memoranda are not yet completed funding commitments.
Washington is pushing its AI security perimeter deep inside the data centre, targeting Chinese-made components that move data between GPUs. The policy may reduce cyber and espionage risks, but it could also raise costs, slow construction and expose a new weakness in America’s AI race as AI scales.
Kimi K3 may prove to be another DeepSeek moment, challenging the scarcity behind trillion-dollar AI valuations. As open intelligence spreads, frontier models may become utilities, while the greater prize shifts to the nations, industries and people bold enough to build with them and share freely.
We are racing to shape our AI future through a new Office of AI and national standards. Yet billions flow into foreign-led data centres while we offer little support for local models or sovereign compute. Without stronger action we risk becoming high-quality hosts rather than true leaders.
Where cybersecurity meets innovation, the CNC team delivers AI and tech breakthroughs for our digital future. We analyze incidents, data, and insights to keep you informed, secure, and ahead. Sign up for free!