Rising AI Bills Push Singapore Companies to Chase ROI Over 'Tokenmaxxing'
Source: The Business Times
For the past two years, Singapore's corporate AI playbook has been simple: throw tokens at the problem and figure out the cost later. That era is ending as companies wake up to the reality that AI without clear ROI is a fast track to budget blowouts, and they're responding by getting surgical about which models to use.
For the past two years, Singapore's corporate AI playbook has been simple: throw tokens at the problem and figure out the cost later. That era is ending. Companies across the island are now waking up to the reality that AI adoption without a clear return on investment is a fast track to budget blowouts — and they're responding by getting surgical about which models to use and when.
The Business Times reports that while the cost of individual AI tokens has fallen sharply — Bain & Co estimates prices halved between 2024 and 2025 — overall AI bills continue to climb as usage expands far beyond initial pilot projects. The paradox is straightforward: cheaper per-unit costs mean nothing when consumption grows exponentially, and Singapore's businesses are now reaching that inflection point where the finance department starts asking hard questions.
This shift comes at a time when Singapore's AI integration into daily work remains paradoxically low even as adoption rates are high — a gap that suggests many companies bought into the technology without building the infrastructure to measure whether it's delivering value. The result is a growing divide between organisations that are treating AI as a genuine productivity lever and those that are simply burning through API credits without a strategy.
The trend mirrors what's happening globally but carries particular weight in Singapore, where the government has aggressively pushed AI adoption through programmes like Smart Nation and the National AI Strategy. Local enterprises are now caught between policy encouragement to go all-in on AI and the real-world pressure of showing returns to stakeholders. The ones that figure out model-to-workload matching — using smaller, cheaper models for routine tasks and reserving frontier models for high-value work — will be the ones that sustain their AI investments through the next cycle.
Why it matters for Singapore: The city-state's AI ambitions depend on more than just adoption numbers — they depend on sustainable deployment. If Singapore's businesses burn out on AI spending without seeing returns, the pushback could stall the ecosystem just as it's gaining momentum. The companies that crack the ROI question now will define how AI scales across the economy, and the playbook they write will matter far beyond their own bottom lines.
