The U.S. hyperscaler capex cycle is the largest coordinated capital deployment in corporate history. Google, Amazon, Microsoft, and Meta plan to allocate approximately $725 billion to capital expenditures in 2026, up roughly 77% from 2025’s $410 billion. Nearly all of this incremental spend is GPUs, custom silicon, data centers, and power infrastructure — not headcount in the traditional sense.
The labor question is not “who gets hired in the U.S.” It is “who captures the engineering and services work that sits around that capital.” And the answer for Indian engineering talent is a genuine bifurcation, not a uniform win or loss.
The Structural Casualty: Traditional IT-Services Labor Arbitrage
The classic Indian IT services model — TCS, Infosys, Wipro, HCLTech — built its scale by billing Western clients for large teams of engineers doing custom development, maintenance, and testing. AI-assisted coding and automation directly compress the number of people needed per client engagement, and the data is unambiguous:
TCS cut approximately 23,460 employees in FY26, the largest headcount reduction among Indian IT majors, as it pivots to an AI-first services model. More than 7,700 senior professionals with over 15 years of experience have lost jobs across the four majors and Cognizant, with roughly 43% moving to other tier-one IT firms and approximately 48% landing in Global Capability Centres. Wipro cut its fresher hiring guidance to 7,500–8,000, well below earlier estimates, with some recruits reporting onboarding delays exceeding seven months. Hiring at India’s big four outsourcers has essentially stalled — the four companies added just 3,910 net staff over a full year, an unusually slow pace by historical standards.
The pyramid-shaped hiring model that made Indian IT firms among the world’s largest employers is structurally shrinking at the base. If you are a mid-career engineer at a services firm doing commodity coding, testing, or maintenance, or a fresher hoping to enter through the old “hire in bulk, train on the job” pipeline, this cycle is working against you.
The Expansion: GCCs, Frontier AI Labs, and Chip Design
Running in parallel is an enormous expansion of high-value engineering work being placed in India directly by the same companies driving the capex boom.
Global Capability Centres — captive engineering and R&D units that multinationals own outright in India, as opposed to outsourcing to a vendor — are projected to cross 510,000 jobs in 2026, a 3.4x increase since 2021, even as broader IT hiring is flat to declining. Nearly two-thirds of new GCC roles in 2026 now require AI, data science, or intelligent automation skills. India now hosts over 2,100 GCCs employing more than 2.3 million professionals and generating over $70 billion in annual revenue. GCCs generally pay a 12–20% salary premium over traditional IT service firms for comparable roles, with the senior-level gap even wider.
Frontier AI labs are establishing direct India operations. Anthropic announced its first India office in Bengaluru, opening in early 2026, with Applied AI Architect and Applied AI Engineer roles posted and heavy investment in Indic-language Claude capabilities. OpenAI is scaling toward approximately 8,000 employees globally by end of 2026 with an India presence. Google DeepMind operates teams in Bengaluru and Hyderabad. This is a genuinely new phenomenon — frontier labs, not just hyperscaler cloud arms, building India-based engineering teams.
Semiconductor and chip design is arguably the most underrated winner category. Indian engineers constitute roughly 20% of the world’s semiconductor designers, working at India centers for Qualcomm, Intel, AMD, Nvidia, and MediaTek. Qualcomm taped out a 2nm chip design entirely from its Bengaluru, Chennai, and Hyderabad centers. AMD is expanding its Bengaluru design center and plans to add 3,000 more engineers by 2028 on top of 6,500 existing employees. This is design-stage engineering feeding directly into the GPUs and custom silicon absorbing the hyperscaler capex.
The Salary Tiering
The compensation picture is sharply tiered, and the gaps within India now often matter more than the India-versus-U.S. gap for any given individual’s outcome:
A senior backend engineer at a mid-size IT services firm might earn ₹18–24 lakh annually. The same title at a GCC in the same city commands ₹35–50 lakh. GCCs and funded product companies pay 40–60% more than services firms at the same experience level. Senior engineers working remotely for U.S. firms from Bengaluru or Hyderabad can earn the equivalent of ₹60–80 lakh. For comparison, a U.S. senior AI engineer commands $220K–$310K base and $340K–$550K total compensation, with frontier-lab specialists going well above that.
Even after adjusting for cost of living, the U.S. premium remains real — a senior Indian AI engineer typically costs 30–60% of a fully-loaded U.S. senior engineer — which is exactly why hyperscalers and labs keep expanding India engineering rather than hiring exclusively in the Bay Area. But within India, the dispersion between a services-firm salary and a GCC or frontier-lab salary is now often as large as part of the India-U.S. gap itself.
The Visa Constraint
The U.S. immigration system is becoming a weaker channel for individual mobility right when U.S. firms need more AI and infrastructure engineering capacity.
The $100,000 H-1B fee — imposed in September 2025 on new H-1B petitions for beneficiaries outside the U.S. — falls disproportionately on Indian tech workers, who account for approximately 71% of all H-1B recipients and over 80% of computer-related H-1B positions. A Massachusetts federal court vacated the fee in June 2026, ruling it an unauthorized tax exceeding executive authority, but then issued an administrative stay effectively reinstating it while the government’s appeal proceeds. The fee remains in effect pending appeal, and the legal status is genuinely unsettled.
The green card backlog is arguably a bigger long-run structural issue. Indian EB-2 applicants face priority dates from around 2013, translating to waits exceeding 12 years, driven purely by the 7% per-country cap. At current processing rates, some estimate new EB-2 PERM filings from India could face 50- to 100-plus-year waits. A Cato Institute analysis estimated over 400,000 Indian applicants could die before receiving a green card under current trends.
The capacity gap is being filled by GCC and India-office expansion instead of import of individuals. That is good for engineers who stay in India and bad for those whose career plan depended on relocating to the U.S.
The Reshuffling
The premium has shifted from “getting to the U.S.” toward “having the specific skills the AI buildout needs, wherever you sit.” India is capturing a disproportionate and growing share of the actual engineering work, just not always via the visa route or the traditional service-company employer.
Engineers with applied AI/ML, MLOps, chip design, or cloud-infrastructure skills — especially at GCCs, frontier lab India offices, or semiconductor design centers — are the clear winners. Mid-career, mid-skill engineers at traditional IT services firms doing commodity development and testing are the clear losers. Anyone whose plan was “H-1B to the U.S. as the default career path” sits in an increasingly ambiguous position — that route is more expensive, legally uncertain, and terminates in a multi-decade green card wait, even as the India-based version of the same work is expanding and paying better than it used to.
India’s IT sector employs approximately 5.8 million people. Slower hiring at the majors ripples across engineering colleges and entire tech cities built around outsourcing. But the high-value segment — GCCs, frontier labs, chip design — is expanding fast enough that the net effect is a reshuffling, not a collapse. The question is not whether Indian engineering talent is a winner or loser in the AI buildout. The question is which segment of Indian engineering talent you are in.