Tech • AI • Robotics • Game
AI Goes Off the Rails… Washington and Brussels Bring Out the Big Guns | Octogone Tech #17
8/10TL;DR
A new wave of tech and AI developments is exposing the fragility of cloud-dependent services, intensifying the race over infrastructure, and sharpening divisions over sovereignty, profitability and control.
KEY POINTS
Google Cloud outage hit Revolut users
A recent Google Cloud outage disrupted Revolut payments, leaving some customers unable to use their cards at critical moments. The incident highlighted a broader vulnerability in digital finance: when banks and fintech platforms rely on a small number of cloud providers, a technical failure can quickly become a consumer-facing crisis. The episode also revived concerns about whether users truly control assets held within app-based financial systems.
AI companies face mounting questions over losses
The economics of frontier AI are under growing scrutiny. Anthropic was described as burning roughly $80 billion a year, while OpenAI had previously been associated with annual losses around $40 billion. Investors are increasingly questioning sky-high valuations, including talk around a possible $2 trillion listing for major AI firms, especially as these companies remain heavily dependent on outside infrastructure and unproven long-term margins.
Cheaper and faster models threaten the current AI race
Pressure is also coming from efficiency gains. A new model associated with Stanford, referred to as Ralous, was presented as running up to 31 times faster, potentially allowing far more users to be served from the same data center footprint. That kind of performance shift could undermine the strategy of firms focused on ever-larger, more expensive frontier systems, while strengthening the case for lighter, more local and more agent-oriented AI.
Local AI is emerging as a strategic alternative
The debate is shifting from raw benchmark supremacy to usability, deployment and ownership. Smaller models running directly on personal computers are increasingly seen as a plausible near-term alternative to cloud-only systems, with expectations that desktop-class hardware could soon handle capabilities closer to today’s top-tier assistants. That prospect would weaken the grip of centralized providers and give users more direct control over data, costs and customization.
Musk’s infrastructure strategy stands apart
Elon Musk was contrasted with loss-making AI rivals through the performance of SpaceX, described as profitable and outperforming expectations. The broader argument is that Musk’s strength lies in vertical integration: controlling infrastructure, computing, hardware and software across companies such as Tesla, SpaceX and xAI. In this view, AI is not a standalone consumer product but a tool embedded inside industrial systems, vehicles, manufacturing and aerospace.
Samsung benefits from the AI memory boom
The AI buildout is also reshaping the semiconductor market. Samsung was cited as posting around 100,000 billion won in revenue, with profits reportedly rising eightfold year on year as demand for memory and chips surged. At the same time, the boom is feeding inflation elsewhere, including a reported 22% drop in smartphone purchases as higher component costs filter through to consumers.
Apple’s foldable push shows both polish and limits
Discussion around Apple’s foldable device, referred to as the iPhone Duo, reflected a familiar split: industrial refinement on one side, software compromises on the other. Early reactions praised the hardware feel and hinge quality, yet criticism focused on black bars, under-optimized apps, restrictive multitasking and doubts over long-term durability. At a reported price near $3,000, expectations for flawless execution are especially high.
German platform drops Mistral for Chinese models
In a notable setback for European AI cohesion, German search engine Ecosia moved away from Mistral and toward Chinese models hosted in German data centers. The decision was framed as a sovereignty issue, but it also underscored fractures inside Europe’s tech ambitions. For supporters of Mistral, the move reinforced the idea that so-called European alignment often breaks down when national industrial interests diverge.
Mistral deepens its scientific role with CERN
Even as it lost a commercial partner, Mistral advanced in research through a new partnership with CERN. The aim is to build a foundation model for physics and particle collisions, extending AI beyond chatbots into scientific modeling. The initiative points to a different path for European AI: less dependent on consumer hype, and more focused on domain-specific tools with strategic value in research and industry.
CONCLUSION
The latest developments suggest the tech sector is entering a more demanding phase, where resilience, infrastructure ownership and practical deployment matter as much as headline innovation. As AI spreads deeper into finance, industry and science, control over the stack may become the decisive advantage.
Ask a question