The most consequential decision in technology this week was not made in a boardroom in Silicon Valley, but in the renegotiation of a thousand smaller contracts in Bengaluru and Hyderabad. India’s IT services giants—TCS, Infosys, Wipro, HCLTech, and Cognizant—are abandoning the industry’s century-old billing model, shifting from charging by the hour to pricing based on performance outcomes 1. Clients, emboldened by the promise of artificial intelligence, are demanding steep price cuts in exchange for productivity gains 1. This is not a minor accounting change; it is the formal admission that the value of human labor in the software industry is now measured against a machine’s output.
The logic is brutal and simple. If an AI model can draft the code, the client asks why they are paying for the human who merely reviews it. Persistent Systems’ CEO has framed this as a necessary evolution 1, but the internal reality for the sector’s engineers is a forced march toward higher utilization and thinner margins. The shift to outcome-based pricing transfers risk from the client to the worker, who must now guarantee results that depend on tools they do not control. This is the organizational culture of tech, stripped of its perks: the employee is no longer a cost-plus line item, but a variable cost to be optimized.
This pressure is not isolated to India. The same week, Amazon quietly raised prices on its Echo, Kindle, and Fire TV lines by up to 60 percent, blaming "significant increases in memory and storage component costs" 12. On the surface, these are unrelated events. But together, they illustrate the new bargain of the AI era: the cost of hardware goes up as the cost of labor goes down. Consumers absorb the former; workers absorb the latter. Amazon’s move is a signal that the cheap-device era is over, while the Indian outsourcing shift signals that the cheap-labor era is also ending—not because wages rose, but because the labor itself is being devalued relative to the algorithm.
The industry’s response to these pressures is to push more automation onto the user. Meta has expanded its AI assistant into a work tool for small businesses, capable of creating presentations and analyzing ad performance 9. Google is adding a chatbot interface to its Discover feed 11. These are not neutral conveniences; they are mechanisms to extract more productivity from the same human hours. The Pope’s warning this week that AI risks becoming "another instrument of ideological or economic colonialism" 4 is a theological framing of a very secular problem: the dependency of poorer nations on the platforms and pricing power of richer ones.
Even the spectacle of Chinese humanoid robots breaking sprint records—a robot running 100 meters in 9.32 seconds 5—reinforces the theme. Unitree’s "Superman" robot can outrun Bolt but struggles to brake 3. The engineering triumph is real, but the inability to decelerate is a metaphor for the entire industry: we are racing forward on capability without a clear mechanism for control or consequence. The backlash against Flock Safety’s license plate readers, where residents are vandalizing cameras to avoid being surveilled 10, is the human attempt to apply the brakes.
The tradeoff is now explicit. For the reader, the question is not whether AI is good or bad, but who pays for the transition. The evidence suggests the bill is split: consumers pay more for hardware, workers accept less for their time, and citizens pay with their attention and privacy. The unresolved question is whether the promised productivity gains will ever be distributed beyond the shareholders of the firms making the cuts. The speed is impressive. The braking mechanism is not yet in sight.