From Silicon to Space: How AI Infrastructure is Racing Beyond Earth's Limits

Today's AI landscape reveals a fascinating paradox: as companies race to build computing infrastructure in space, earthbound data centres face unprecedented geopolitical threats that could reshape the entire industry.

The Final Frontier for AI Computing

The race to space-based AI infrastructure has officially begun, with Cisco CEO Chuck Robbins declaring his company will "absolutely" build data centres in space. Robbins cites unlimited solar power and the elimination of terrestrial opposition as key advantages, revealing that Cisco teams began space data centre preparations 2-3 months ago.

This isn't just tech industry bravado. Space-based computing addresses real constraints facing AI infrastructure: energy costs, cooling requirements, and increasingly vocal community opposition to ground-based facilities. By eliminating atmospheric cooling systems and tapping into continuous solar energy, orbital data centres could solve the sustainability equation that has plagued the AI boom.

The timing coincides perfectly with Elon Musk's million-satellite constellation plans, creating a potential symbiotic relationship between satellite networks and space-based computing power. For organisations planning long-term AI strategies, this shift suggests that tomorrow's most powerful AI systems may literally be out of this world—with profound implications for latency, data sovereignty, and access equity.

While Robbins positions this as being on the "leading edge" rather than "bleeding edge," the move signals a fundamental rethinking of where critical AI infrastructure can and should be located. The question isn't whether space-based AI computing will happen, but how quickly it will become competitive with terrestrial alternatives.

When Geopolitics Meets AI Infrastructure

Iran's Islamic Revolutionary Guard Corps has escalated tensions by threatening to destroy OpenAI's $30 billion Stargate data centre in Abu Dhabi, marking a dangerous new chapter in the intersection of geopolitics and AI infrastructure. The threat, delivered through Iranian state media, specifically targets the massive facility if the US attacks Iran's power plants.

This isn't an isolated incident. Iran has already struck AWS data centres in Bahrain and Oracle facilities in Dubai, demonstrating that critical AI infrastructure has become a legitimate target in international conflicts. The attacks represent a significant escalation where digital infrastructure is treated as strategic military assets.

The vulnerability extends beyond individual facilities to the broader $500 billion Stargate initiative involving Oracle and other major partners. For organisations building AI strategies around cloud infrastructure, this reality check is sobering: your most critical AI workloads may be running in facilities that are now explicit targets of nation-state actors.

The geopolitical risk isn't theoretical—it's actively reshaping how companies think about AI infrastructure resilience. Multi-region deployment strategies that once focused on latency and compliance now must consider the possibility of entire data centre regions becoming inaccessible due to military action. This new reality may accelerate both the move to space-based computing and the development of more distributed, resilient AI architectures.

The Democratisation of Strategic Intelligence

Traditional consulting is facing an AI-powered disruption as Indian startup Rocket launches a platform that generates McKinsey-style reports for $250-350 per month, compared to thousands for traditional consulting engagements. With 1.5 million users and $15M in funding, Rocket positions itself as addressing the strategic "what to build" gap as AI coding tools commoditise the "how to build" aspect.

This trend extends into investment strategy, where OpenAI alumni have launched Zero Shot, a $100M venture capital fund leveraging insider knowledge to identify promising AI startups while avoiding overhyped areas. The fund reflects how AI expertise is becoming a distinct competitive advantage in investment decision-making.

Meanwhile, family offices are increasingly bypassing traditional VCs to invest directly in AI startups, with 83% citing AI as a strategic priority. This shift toward direct participation suggests that traditional gatekeepers in both consulting and venture capital are being disintermediated by AI-powered alternatives.

For organisations, this democratisation of strategic intelligence represents both opportunity and risk. While access to McKinsey-quality analysis at consumer prices could level the playing field, it also means competitive advantages from exclusive strategic insights may erode faster than ever. The winners will be those who can combine AI-generated intelligence with uniquely human judgment and execution capabilities.

The Technical Reality Check

Behind the hype, AI practitioners are grappling with fundamental performance issues that reveal the gap between marketing promises and engineering reality. A detailed analysis of Claude Code shows a 70% drop in research-to-edit ratio since February, coinciding precisely with Anthropic's rollout of "thinking content redaction." The regression demonstrates how seemingly minor API changes can devastate complex engineering workflows.

The issue highlights a broader tension in AI development: the pressure to optimise for cost and speed often conflicts with the deep reasoning required for sophisticated tasks. When Anthropic reduced Claude's "thinking tokens"—the extended reasoning process that happens before generating responses—the model's ability to handle multi-file engineering projects collapsed.

This technical reality check comes as Meta AI releases EUPE, a compact vision encoder under 100M parameters that matches specialist models, proving that efficiency gains don't always require performance sacrifices. EUPE's success suggests that the future lies not in simply scaling up models, but in smarter architectures that maintain capability while reducing computational requirements.

For organisations deploying AI in production, these developments underscore the importance of rigorous testing and performance monitoring. The most sophisticated AI systems can experience dramatic capability regressions from seemingly minor updates, making continuous evaluation and fallback strategies essential for mission-critical applications.

Quick Hits