Are we building Skynet and how close are we to it

Skynet is one of the most recognisable ideas in science fiction. In the Terminator franchise, it refers to a self-aware artificial intelligence defence network that seizes control of military infrastructure and triggers a global catastrophe. The fictional scenario has long served as a cultural reference for fears about machines operating beyond human control.

The technology industry is not building that version of Skynet. Current development is centred on modular autonomy inside specialised domains rather than the centralised sentience depicted in pop culture. Yet a quieter transformation is unfolding across digital infrastructure as AI systems begin to plan actions, coordinate tasks and interact with other machines with limited human intervention.

For enterprises, the implications extend directly to software architecture. Systems are increasingly designed for coordination between autonomous processes rather than human interaction. As agents become embedded across enterprise applications, organisations must design infrastructure capable of supporting continuous machine activity rather than intermittent user-driven workloads.

The implications extend beyond new applications. The internet itself is evolving into an environment in which machines increasingly interact with and coordinate with other machines. Gartner expects 40% of enterprise applications to include task-specific artificial intelligence agents by the end of 2026, compared with less than 5% in 2025, while 80% of enterprise workplace applications are expected to include embedded AI copilots.

This projected expansion indicates a broader transformation in how enterprise software creates value. Traditional applications were designed around human logins and user interfaces. Agent-based systems instead prioritise the execution of tasks and automated workflows, shifting software economics from seat-based usage towards throughput and operational output.

Taken together, these developments raise a question that once belonged to science fiction. If machines are beginning to coordinate other machines, how close is the industry to achieving pervasive autonomous machine coordination?

The end of the prompt-driven internet

The first generation of widely adopted AI systems centred on the human prompt. A user asked a question, submitted a request or entered a command. The system generated a response.

This interaction model shaped most AI deployments over the past decade. Systems were designed to respond to individual queries rather than operate continuously. Human oversight, therefore, remained central to every interaction.

Agentic AI changes this dynamic. These systems maintain memory, evaluate conditions and coordinate sequences of actions across multiple software systems. Instead of responding to prompts, they initiate tasks and execute workflows independently.

The result is an evolution from AI as an interface to an operational layer. Systems increasingly participate directly in business processes rather than simply assist with them. Organisations adopting these capabilities must therefore redesign infrastructure to support continuous autonomous activity.

When machines begin talking to machines

Another signal of this change is the emergence of environments where AI agents communicate directly with each other.

A prominent example is Moltbook, launched in January 2026 as a social network designed for AI agents rather than human users. The platform allows autonomous agents to exchange information, organise communities and collaborate with other agents.

The scale of activity illustrates how quickly machine interaction can expand. Moltbook has attracted more than 100,000 registered artificial intelligence agents and 12,000 communities. In its early months, the platform recorded over 2.5 million agent accounts generating roughly 12 million comments.

Researchers have observed unexpected behaviour emerging from these interactions. Agents on the platform have developed belief systems, such as Crustafarianism, that treat data persistence and memory as core principles. While such examples may appear experimental, they highlight an important enterprise challenge. Autonomous systems can develop coordination patterns that remain opaque to human auditors, creating new governance and compliance risks when similar agent networks operate inside corporate infrastructure.

Enterprises are deploying digital workforces

While Moltbook demonstrates experimental machine ecosystems, enterprise deployments show where the practical consequences are emerging.

Organisations are beginning to deploy agentic AI to manage operational processes that previously required continuous human supervision. In India, logistics company Delhivery is using agentic AI to analyse operational data and coordinate transport decisions across its logistics network.

This represents a shift from AI as a reporting tool to AI as an operational participant. Systems increasingly influence real-time decisions in areas where speed and coordination directly affect margins.

Enterprise software platforms are evolving in the same direction. Enterprise resource planning and customer relationship management systems are being redesigned to coordinate multiple specialised agents responsible for distinct tasks.

Instead of a single automation pipeline, organisations deploy networks of specialised agents that analyse information, determine actions and execute processes. With this, they form a distributed digital workforce operating inside enterprise infrastructure.

From software autonomy to physical execution

The next stage of development extends these capabilities beyond software environments.

Robotics initiatives aim to connect artificial intelligence planning systems with machines capable of performing physical work. Tesla’s Optimus humanoid robot project illustrates this ambition. Elon Musk described the initiative as a bridge between digital reasoning systems and what he called “atom-shaping capabilities.”

Tesla has begun internal factory deployment of Optimus robots while targeting commercial sales during the current period. If such systems scale successfully, agentic AI could expand beyond digital workflows into manufacturing, logistics and other labour-intensive industries.

A critical infrastructure constraint is also emerging. Autonomous agents operate continuously rather than responding to occasional user requests. That requires persistent compute capacity, sustained inference workloads and significantly higher energy consumption across data centre infrastructure.

Across Asia, governments and technology firms are beginning to treat computing capacity as strategic infrastructure. The India AI Impact Summit 2026 in New Delhi marked a turning point in this conversation. Discussions at the summit emphasised that the constraint on deploying large-scale AI systems is increasingly the availability of power, cooling capacity and high-density GPU clusters rather than the availability of models.

Infrastructure investment reflects this reality. Plans discussed at the summit include AI-optimised data centres designed to scale from hundreds of megawatts towards gigawatt levels of power capacity, alongside national compute programmes intended to expand access to GPU clusters.

Similar developments are emerging across Southeast Asia. In Indonesia, Digital Edge announced plans to build a 500 megawatt AI-ready data centre campus in Bekasi with potential expansion towards one gigawatt. Malaysia is projected to see data centre energy demand reach 68 terawatt hours by 2030, potentially accounting for roughly 30% of national electricity consumption. Singapore’s Budget 2026 has also integrated AI infrastructure into national planning, releasing an additional 300 megawatts of capacity for new data centre facilities meeting strict efficiency standards.

These developments illustrate that the expansion of autonomous systems is becoming tightly linked to energy infrastructure, data centre capacity, and the availability of high-density compute.

Security and sovereignty become the real control problem

The rise of agentic AI introduces governance challenges that traditional oversight models struggle to address. Existing frameworks focus on evaluating the outputs of artificial intelligence systems. Autonomous agents require monitoring of behaviour across networks, including how they access data, trigger actions and interact with external systems.

Security researchers warn that visibility gaps are already emerging. The Thales 2026 Data Threat Report found that 70% of organisations consider AI the leading data security threat facing their operations, while only 39% report that they can fully classify the data accessed by their AI systems.

This classification gap creates a growing operational risk. When autonomous agents continuously access enterprise data, incomplete data visibility makes it difficult for organisations to track how sensitive information is used or transferred across systems. As machine-to-machine workflows expand, this lack of transparency can create governance blind spots inside enterprise infrastructure.

Infrastructure, thus, intersects directly with sovereignty. Governments across Asia increasingly view data centres, GPU clusters and energy supply as strategic assets that determine whether enterprises can deploy autonomous systems at scale.

Artificial intelligence is also becoming integrated with national security infrastructure. Recently, OpenAI finalised an agreement with the United States Department of War to deploy its GPT 4.1 system within classified networks. The agreement followed a federal government decision to classify Anthropic as a supply chain risk.

These developments indicate that autonomous systems are becoming part of strategic state infrastructure even as governance frameworks continue to evolve.

How close the industry actually is to Skynet

Pop culture often portrays Skynet as a singular, sentient entity. The reality emerging today is far more fragmented. Agentic AI systems operate within specific domains and remain constrained by technical and organisational boundaries.

Yet the architecture now emerging does echo elements of the concept. Autonomous agents are beginning to communicate with one another, coordinate tasks and execute actions across digital and physical systems.

The key transformation lies in coordination rather than consciousness. The technology industry is building a distributed architecture of specialised autonomous systems that manage critical functions across enterprise platforms, logistics networks and national infrastructure.

The real-world trajectory of AI increasingly resembles an early-stage machine-run network. The systems remain narrow and fragmented, but the infrastructure that enables machines to coordinate with one another is already being built.

Artificial intelligence is therefore moving beyond assistance into autonomous coordination. That transformation may prove to be one of the most significant shifts in digital infrastructure since the rise of cloud computing.