Google CEO called out 'biggest AI budget problem' of companies world over from IO stage with a solution

Google CEO Sundar Pichai shifted the AI speech from to economics astatine this years’s Google I/O conference. Pichai warned that the companies astir the satellite are blowing done their yearly AI budgets by May owed to runaway token usage.

Pichai said the accelerated emergence of AI agents has created unprecedented costs for enterprises. “Companies are already blowing done their yearly token budgets and it’s lone May,” helium noted, referring to the billions of tokens consumed by AI systems each month. Pichai said the accelerated emergence of AI agents has created unprecedented costs for enterprises. “Companies are already blowing done their yearly token budgets and it’s lone May,” helium noted, referring to the billions of tokens consumed by AI systems each month.

Suggesting a solution Sundar Pichai announced Gemini 3.5 Flash

Pichai unveiled Gemini 3.5 Flash, a exemplary which is designed to rival frontier offerings portion cutting costs. He said companies could prevention implicit $1 cardinal annually if they shifted 80% of workloads to a premix of Flash and different frontier models. “If companies utilized a premix of Flash and different frontier models they could prevention a batch of money. To enactment this successful perspective, apical companies are processing astir 1 trillion tokens a day. If they shifted 80% of their workloads from different frontier models to 3.5 Flash, they’d prevention implicit $1 cardinal dollars annually.

That is existent savings they tin determination backmost into their company,” added Pichai.

What is Gemini 3.5 Flash model

Google unveiled Gemini 3.5 Flash, its latest AI model, astatine the I/O 2026 developers conference. The institution has positioned this exemplary arsenic the archetypal merchandise successful the caller Gemini 3.5 family. The institution said the AI exemplary is designed to present improved show for coding, agentic AI tasks and multimodal knowing portion maintaining the faster effect speeds associated with the Flash series.According to Google, Gemini 3.5 Flash is present disposable globally done the Gemini app, AI Mode successful Search, Google AI Studio, Android Studio and endeavor platforms. The exemplary besides powers caller AI experiences crossed Google products, including the upcoming Gemini Spark idiosyncratic AI agent.The institution said Gemini 3.5 Flash performs amended than Gemini 3.1 Pro crossed respective coding and agentic benchmarks, including Terminal-Bench 2.1, GDPval-AA and MCP Atlas.

Google added that the exemplary is susceptible of handling long-horizon workflows, enabling it to execute multi-step tasks specified arsenic exertion development, codification attraction and papers preparation.Google besides highlighted improvements successful multimodal capabilities, stating that Gemini 3.5 Flash tin make interactive web interfaces, graphics and animations portion supporting much analyzable reasoning tasks. The exemplary is designed to enactment with Google’s updated Antigravity platform, allowing aggregate AI subagents to collaborate connected larger workflows.

Why Google has an edge

Google’s vantage lies successful owning the afloat stack chips, information centers, cloud, models, and applications. Analysts estimation Google pays 50% to 75% little for AI compute than rivals acknowledgment to its customized TPU chips and nonstop sourcing. By contrast, competitors similar OpenAI trust connected Microsoft, Oracle, and Nvidia, paying margins astatine each furniture of infrastructure.