Google can afford to offer its LLM for free, creating immense pricing pressure on competitors like OpenAI. This strategy aims to eliminate competition by making their business models unprofitable, securing a monopoly for Google before it begins to monetize.
Google's DNA is rooted in the high-margin search business. This cultural bias, combined with public market pressure, makes it difficult to pursue a long-term, zero-profit "bleed out" strategy for Gemini, even if it could secure a monopoly.
A user arriving from ChatGPT has likely already used the AI to research, compare options, and make a decision. This pre-qualification process means the traffic is extremely high-intent, leading to conversion rates (12% for one brand) that are an order of magnitude higher than typical benchmarks.
While purpose-built chips (ASICs) like Google's TPU are efficient, the AI industry is still in an early, experimental phase. GPUs offer the programmability and flexibility needed to develop new algorithms, as ASICs risk being hard-coded for models that quickly become obsolete.
While OpenAI's projected losses dwarf those of past tech giants, the strategic goal is similar to Uber's: spend aggressively to achieve market dominance. If OpenAI becomes the definitive "front door to AI," the enormous upfront investment could be justified by the value of that monopoly position.
OpenAI CEO Sam Altman's move to partner with a rocket company is a strategic play to solve the growing energy, water, and political problems of massive, earth-based data centers. Moving AI compute to space could bypass these terrestrial limitations, despite public skepticism.