Google Companies Seeing

AI's Broken Economics and the Information Wars Beneath Them

Goldman Sachs released a sobering assessment this week concluding that AI economics are actually worse now than they were two years ago — a striking indictment of an industry that has absorbed hundreds of billions in investment since the ChatGPT boom. The finding suggests either that the technology is not delivering promised returns or that implementation costs have proven far higher than anticipated.

Against that backdrop, Google completed its Gemini 'Thinking Levels' rollout across all platforms, representing the company's bid to match OpenAI's reasoning capabilities. Separately, Google launched its encoder-free Gemma 4 12B model, designed to process text, images, and audio on consumer laptops equipped with as little as 16 gigabytes of memory — a move toward edge computing that would reduce dependence on cloud infrastructure and allow local data processing.

TSMC Chief Executive C.C. Wei told shareholders that chip supply will not meet AI demand for years, even with new US manufacturing capacity coming online. The company projected over 30 percent sales growth for the year but said it still cannot fulfill surging AI chip orders — a physical bottleneck that limits AI deployment regardless of software advances. Nvidia's Jensen Huang was separately reported to be pitching what he called 'insane' AI returns to billionaire family offices, though photonics stocks pulled back after a Huang-fueled rally, suggesting investor enthusiasm may have outpaced realistic growth projections.

A new form of information pollution has emerged alongside these hardware constraints. Companies are reportedly spamming Reddit in order to manipulate responses from ChatGPT and Google AI systems. Because AI models are trained on Reddit data, systematic manipulation of that content can compromise AI outputs without users being aware — a subtle but significant form of information warfare.

In the legal arena, a judge ordered Elon Musk to hand over Tesla and SpaceX emails in the xAI lawsuit, raising the possibility of shareholder fraud if company resources were improperly used. A separate legal dispute saw xAI ask courts to reveal the identities of individuals who are plaintiffs in a Grok deepfake case, while plaintiffs argued that disclosure would cause further harm and deter future victims from coming forward.