Budget 2025-26 AI Scorecard: 8 Promises Graded — Verdict C+

The compute pillar is the Budget’s strongest structural investment. The IndiaAI Compute Portal makes high-end GPUs accessible to Indian startups and research institutions at subsidised rates — directly addressing the compute cost barrier that has historically forced Indian AI companies to depend on foreign cloud providers. The initial procurement of 10,000 GPUs is a meaningful start, but analysis from NITI Aayog’s own technical assessments suggests full domestic AI sovereignty for BharatGen-class models requires a 3× scale-up over three years. Grade B rather than A because the deployment timeline is 24–36 months, not immediate.

The FutureSkills pillar targets 1 million AI-certified workers by FY2027. The number sounds significant until you compare it to the displacement projections from our Hourglass analysis: 8–12 million workers in AI-exposed white-collar roles facing structural headwinds by the same date. At 1 million certifications against 8–12 million at-risk workers, the coverage ratio is 1-in-8 at best. The programme is also voluntary and pull-based — workers must seek it out. South Korea’s equivalent programme is push-based and employer-funded. That structural difference explains why the grade cannot be higher than C−.

The BharatGen and AIKosh investments are the Budget’s most strategically differentiated move. By investing in open-source, publicly owned, multilingual foundational models — rather than licensing proprietary foreign AI — the government ensures India’s enterprises and 490 million informal workers can build on India’s own stack without paying licensing fees to foreign monopolies. This directly addresses the critical Scenario A vs. Scenario B fork we identified in the Two Indias analysis. The B+ rather than A reflects that deployment timelines for vernacular voice-AI at rural scale remain 18–24 months out.

AgriStack integration with voice-AI is one of the most powerful displacement mitigation tools available to India — it converts rural workers from passive subsistence farmers into active AI-augmented agritech advisors, creating real income growth without requiring urban migration. But ₹500 crore against a 120 million worker target works out to ₹417 per person — enough to build the digital backbone but insufficient for the multilingual voice-AI deployment, handset subsidy, and last-mile connectivity that make the system usable in Banda, Gorakhpur, or Betul. The Budget commitment is real, the scale is not.

The PLI expansion for electronics manufacturing is not a direct AI-workforce measure, but it is a strategically smart displacement hedge. By creating semi-technical and assembly roles in hardware manufacturing, it absorbs some portion of workers who would otherwise compete for the same narrowing white-collar entry-level jobs. It also builds India’s hardware manufacturing base — reducing dependency on Chinese electronics imports and creating domestic capacity for the compute infrastructure BharatGen needs. The B− reflects that the displacement offset is real but modest (approximately 6% of at-risk workers), and the jobs created are not AI-adjacent in the way FutureSkills roles are.

The e-Shram portal was extended with a minor budget enhancement, but the Budget did not enact the one measure that would have transformed it from a database into a genuine safety net: mandatory platform contributions. The Social Security Code (2020) has been law for six years. Gig platforms operating in India still contribute nothing to workers’ health insurance or pension funds. As AI displacement accelerates and more white-collar workers transition to gig work, this gap becomes structurally catastrophic. The D rather than F reflects that the portal itself was funded and expanded — but the architectural change was absent.

This is the Budget’s most significant failure, and the one with the largest downstream consequences. The Economic Survey 2024–25 explicitly flagged the need to link AI software depreciation deductions to corporate employment retention rates. Finance Minister Sitharaman’s Budget speech did not address it. The result: companies deploying AI to mass-displace workers can still write off 100% of the software cost as capital expenditure, with zero requirement to transition, retrain, or retain any portion of the displaced workforce. This is the single biggest missed opportunity of Budget 2025-26. South Korea enacted the equivalent mechanism in their 2025 National AI Strategy. India deferred it entirely.

The Budget referenced the National Education Policy 2020 framework and digital skills integration in broad terms, but no allocation was made specifically for the AI-era curriculum overhaul that high-displacement states like UP, Bihar, and MP urgently need. The class of 2032 is in school today. If state board curricula continue to teach rote Python syntax memorisation while LLMs perform that function natively, India will produce a generation of graduates with credentials that have no market value. The D+ rather than F reflects that the NEP 2020 framework, if properly implemented, does create a pathway — but Budget 2025-26 provided no new funding or mandate to accelerate it.