Tokenmaxxing isn't an AI strategy
The Register
The article examines the true cost of AI and argues that focusing on token spending as a performance metric is misguided; specific token prices vary widely depending on hardware, utilization rates, and other factors, with Anthropic charging $5/million tokens for input and $25/million for output on Opus 4.7, while companies often deploy expensive AI projects without clear business objectives or ROI measurement. Rather than optimizing for token consumption, organizations should first define business outcomes and desired results before investing in AI systems, as research shows only 15 percent of AI prototypes reach production without strategic planning versus 45-50 percent with deliberate goal-setting. The piece argues that token spending correlates poorly with actual productivity and that companies should prioritize understanding why they need AI and what they want to accomplish rather than simply increasing token usage.
Why it matters
Before checking AI's price tag, see whether it fits
