Boasting about using billions of tokens or writing thousands of prompts is no longer enough to land a job in the artificial intelligence (AI) space. If anything, it could be a red flag for cost-conscious, outcome-focussed startups unless candidates can show how generative AI could help build products faster, cut costs, improve customer experience or boost productivity, founders and recruiters said.
“When we hire folks, the ones who are mentioning: ‘Oh, I max out these many tokens; I’ve used 5 billion tokens or 10 billion tokens in the last one week or three days or whatever'—that's a red flag for me if you're not giving me an ROI (return on investment)," said Additi Upadhyay, co-founder of AI startup Noveum AI. "I think it's been a long time that we have moved to value since we were seeing larger companies cutting down on their AI budgets.”
The shift reflects a broader evolution in enterprise AI adoption—from ‘token maxing’ to ‘value maxing’.
During the initial wave of generative AI adoption, organizations often measured success by the volume of AI usage, reflected in billions of tokens consumed. Tokens are the basic units of text that AI models process, and because most AI providers charge based on token usage, they became a proxy for how extensively companies were using AI.
As AI becomes embedded in core business operations and inference costs come under greater scrutiny, companies are increasingly shifting towards ‘value maxing’, where success is measured by the business outcomes AI delivers rather than the amount of AI consumed.
Experts say this mindset shift is influencing not only how companies hire AI talent but also how they assess existing employees and manage AI spending. Internal reviews are increasingly focused on whether AI usage is delivering measurable business outcomes rather than simply tracking token consumption.
“We're also seeing hiring mandates evolve. Instead of asking whether someone has experience with a particular AI platform, every one in two employers now wants evidence of outcomes: how AI improved workflows, reduced turnaround time or enabled better business decisions," said Neelabh Shukla, chief business officer of Careernet, a talent solutions and recruitment provider. "Technical proficiency remains important, but it's increasingly being evaluated alongside business acumen, critical thinking and the ability to apply AI with purpose.”
Earlier, conversations centered around access to the latest models and how extensively teams were using them. "Today, founders and investors are asking a different question: what measurable business value is AI creating? We're seeing startups as well as enterprises increasingly evaluate AI through metrics like cost-per-outcome (CPO), return on AI investment (ROAI), and time-to-outcome (TTO), rather than simply infrastructure or token consumption. AI is becoming part of the unit economics of a business, and it's a healthy shift," said Sunil Nehra, chief executive officer, IT staffing at FirstMeridian Business Services Ltd.
For instance, recruiters increasingly want to understand whether a candidate can identify the most efficient model for a given task, rather than defaulting to the largest frontier model. Simpler enterprise use cases such as document translation and information extraction can often be handled by smaller models with 12 billion to 70 billion parameters, delivering higher throughput and lower inference costs.
Assessing candidates
Candidate assessments are also now more business-focused. “AI recruitment is moving beyond evaluating technical skills to assessing a candidate’s ability to deliver measurable business outcomes," said Nehra of FirstMeridian. “Interviews today increasingly focus on how AI solutions improved productivity, reduced costs, enhanced customer experience, or drove revenue, not just on the technologies used.”
“As enterprise AI adoption matures, hiring is increasingly favouring professionals who can bridge technology with business value,” Nehra added.
Responding to the market demand, successful candidates' resumes are highlighting skills that could help firms reduce their AI costs: tracking AI spending (cost allocation tagging), reusing previous AI outputs (caching) and optimizing token usage, hiring industry experts said.
“We're also seeing this shift reflected in how candidates position themselves. Rather than emphasizing the number of AI models they've worked with or the volume of AI usage, stronger applications and CVs are increasingly centred around business impact," said Ballurkar of Experis. "Candidates are highlighting how they have used AI to reduce turnaround times, improve operational efficiency, enhance customer experiences or enable better business decisions.”
Economics matters
The shift is largely being driven by economics. As AI becomes deeply embedded in products and workflows, organizations are paying closer attention to their AI spending and token budgets.
On 7 July, Mint had reported on how while Indian firms ramp up AI investments, their boards are increasingly demanding accountability on its returns and risks. Earlier that week, Mint also reported how India’s largest information technology giants are now aggressively tracking their token bills, after having first flagged the phenomenon of tokenmaxing, or excessive AI token consumption, in January.
Uber said it exhausted its AI coding budget for 2026 within the first four months after engineers rapidly adopted Anthropic's Claude Code. The company has since emphasized the need to evaluate AI spending more closely, prompting executives to question whether rising token consumption was translating into proportional business value.
While large technology companies can still afford to experiment at scale, startups argue that they operate under much tighter capital constraints, making AI efficiency even more critical.
"Nvidia CEO Jensen Huang stated that he would be 'deeply alarmed' if a $500,000-a-year engineer or AI researcher doesn't use at least $250,000 in AI tokens annually. This is good for Nvidia, but not necessarily for everyone else," said Srinivas Varadarajan, co-founder and CEO of deeptech startup Vigyanlabs.
Talent crunch
The shift in AI hiring comes even as companies grapple with shortage of the kind of talent they are seeking. According to ManpowerGroup's Global Talent Shortage Survey 2026, 82% of employers report difficulty finding the skilled talent they need, with AI Model & Application Development (39%) and AI Literacy (38%) emerging among the hardest capabilities to hire.
“This tells us that organizations are becoming far more selective. They're looking beyond technical proficiency to professionals who can combine AI expertise with commercial thinking, critical reasoning and human judgement to deliver measurable business outcomes,” said Sanju Ballurkar, president, Experis, ManpowerGroup India. “Ultimately, AI is a tool. What differentiates professionals is their ability to apply that tool in ways that create measurable value for the business.”