On April 28, 2026, Google broke ground on its largest investment in India. The ceremony in Tarluvada, Visakhapatnam, was celebrated as a national milestone—a $15 billion commitment to digital infrastructure and artificial intelligence. What went unmentioned in the official statements was this: the land beneath the ceremonial sod had been redistributed to Dalit families less than fifty years earlier as partial redress for centuries of exclusion. Now the state was acquiring it back.
In Tarluvada, 200 of the 480 acres targeted for the data center belong to Dalit farmers. Pyla Kondamma, a former village council head and farmer, has publicly refused government pressure to sell. "We are not afraid," she said. "Even if they kill us, we will not give it away." Her defiance names what official language obscures: this is not development. This is dispossession wearing the mask of development.
The Tarluvada acquisition is not an anomaly. It is the latest chapter in a 75-year narrative in which successive regimes of capital have dismantled Dalit land security using the same institutional machinery—law, bureaucracy, the rhetoric of progress—that once, briefly, promised to secure it. Understanding this pattern is essential because it reveals something the digital economy wants hidden: that caste-targeted dispossession is structural, not incidental, to how India's peripheral lands are converted into reservoirs of accumulated capital.
Start with what was supposed to be liberation. In 1951, India's Constitution abolished zamindari—the feudal intermediary system that had extracted agricultural surplus for centuries. The reform was radical in intent. Landlordism would end. Land would flow to cultivators. Dalit agricultural laborers, barred from ownership for generations, would become proprietors. B.R. Ambedkar, who had written extensively on how land monopoly anchored the caste system itself, saw in zamindari abolition the possibility of material emancipation.
But the machinery of abolition contained loopholes that ensured caste-based land concentration persisted. The vagueness of "personal cultivation"—what counted as land the zamindar could keep—allowed landlords to evict tenant farmers and retain vast holdings. Upper-caste elites captured the reform process locally. Zamindari abolition loopholes allowed dominant castes to retain control. By the early 1960s, Dalit landlessness remained acute. Ambedkar's prescient warning proved accurate: zamindari abolition alone was insufficient against the monopolization of land by dominant castes, which constituted the material foundation of caste hierarchy.
In the 1970s, Andhra Pradesh attempted to close what zamindari abolition had left open. The state launched a land redistribution program, allocating approximately two acres per household to landless Dalit families. Tarluvada received these allocations, formalized through government deeds. The moment was fragile but real: a second chance at the constitutional promise that Dalit communities might own the land they worked. By 2006, however, a state review found that Scheduled Castes, comprising 16 percent of the population, controlled only 7.5 percent of operated agricultural land. The gains were being eroded.
The turn came with neoliberal development. Beginning in the 2000s, Special Economic Zone projects claimed vast tracts of land for factories, ports, and export zones. These were presented as engines of growth. What they generated, in practice, was a new mechanism of caste-targeted dispossession. Research on SEZ land acquisitions shows that Dalit and Adivasi communities bore a disproportionate share of displacement. In comparative ethnographic studies, indebtedness increased more than 20-fold for Scheduled Caste and Scheduled Tribe households following land acquisition, versus only 1.6 times for other groups. Compensation rates paid to lower-caste farmers averaged approximately half those paid to upper-caste holders. The state apparatus, nominally neutral, functioned as a caste-selective machine.
What has changed from zamindari abolition to SEZ to AI hub is not the underlying logic but the justification and the intensity of technological mediation. Each regime rebrands coercion as opportunity. Land is not seized; it is "acquired" for the "public purpose" of development. The affected are not dispossessed; they are offered compensation and promises of new employment. Digital systems make the theft more efficient. When Andhra Pradesh launched its digitized land records portal, algorithmic errors stripped Dalit and tribal farmers of land claims, erasing cultivation history and falsifying ownership records. The caste mechanism becomes invisible behind code.
The Tarluvada acquisition exemplifies this pattern. The state is acquiring Dalit land while leaving dominant-caste holdings untouched. This selective targeting is not coincidental—it reflects what might be called the operation of neutral procedure as a caste mechanism. Land acquisition statutes make no mention of caste. They are facially equal. Yet in practice, only certain communities face pressure to sell. Only certain compensation rates are offered. Only certain lands are deemed necessary for the public purpose. Administrative procedure, stripped of any explicit caste language, functions as a caste technology.
The deeper logic emerges when we consider what tech capital requires: cheap, peripheral land and massive flows of water and power in water-stressed regions. Data centers demand these resources at scale. The margins of India—areas where Dalit and tribal communities have won fragile claims to land—offer exactly what capital seeks. The Visakhapatnam region is water-stressed. Yet the data center will consume enormous volumes of subsidized water diverted from agricultural use. Coal plants are being reopened to supply power. Environmental toxicity concentrates in the same peripheral zones where dispossessed communities remain.
The scale of this dispossession matters. Nationally, 68 percent of Dalit households own no agricultural land. In Visakhapatnam district, the figure is even bleaker. The 1970s redistribution was, in historical terms, a rare moment when the state allocated land to the landless. What is being reversed now is not merely property but a decades-long effort to translate the constitutional promise of equality into material conditions. It is dispossession of dispossession-prevention.
Economist Sukhadeo Thorat has documented extensively how caste-based economic exclusion operates through systematic discrimination in land ownership, housing access, and credit availability. His research shows that SC households face screening mechanisms in property markets that exclude them regardless of economic capacity. What we are witnessing in Tarluvada is the state itself deploying these mechanisms at scale. The bureaucratic apparatus that once promised to redistribute land has become the instrument of its reconcentration.
Pyla Kondamma's refusal matters not because it will stop the acquisition—capital and the state are too aligned for that—but because it names what must be seen clearly: this is theft. Not development. Not progress. Theft sanctioned by law, rationalized by growth narrative, mediated by digital systems, and executed with the backing of a state that once, briefly, promised something different.
The question that haunts the Dalit constitutional moment is whether emancipation through property is possible, or whether every land redistribution will be reversed when it conflicts with capital accumulation. Tarluvada suggests the latter. What the state gave in the 1970s, it now takes back in 2026. The apparatus of dispossession has simply learned to speak the language of the digital age. Until that apparatus itself is dismantled—until the machinery of the state is fundamentally reoriented toward Dalit liberation rather than capital accumulation—every moment of Dalit land security will remain provisional, a temporary reprieve before the next expropriation.