Microsoft Reaches Carbon Deal with Indian Varaha, But It Exposes the Growing Gap Between Big Tech’s Climate Promises and the AI Emissions Reality

Microsoft has reached an agreement to buy more than 100,000 tons of carbon dioxide removal credits from Indian climate startup Varaha, a deal believed to signal how the global race to scale artificial intelligence is reshaping climate strategies, carbon markets, and even rural farming systems far from Silicon Valley.

At its core, the deal, which is best understood not as a standalone sustainability move, links two very different pressures. On one side is Microsoft’s rapidly expanding cloud and AI business, which has pushed energy demand and emissions sharply higher. On the other hand, India’s long-running struggle with agricultural waste burning is an environmental and public health issue rooted in how millions of smallholder farmers manage crop residues after harvest.

According to TechCrunch, Varaha’s project seeks to connect those worlds by turning cotton stalk waste into biochar, a charcoal-like material that stores carbon in soils for long periods. Instead of being burned in open fields, releasing carbon dioxide and particulate matter into the atmosphere, the waste is processed in industrial reactors. The resulting biochar is then returned to farms, where it can improve soil quality and reduce the need for chemical fertilizers.

Microsoft will purchase the carbon removal credits generated by this process over the next three years, through 2029. The initial phase will focus on Maharashtra, one of India’s major cotton-producing states, and involve roughly 40,000 to 45,000 smallholder farmers. Over time, the project will expand across India’s cotton belt, supported by 18 industrial biochar reactors expected to operate for 15 years.

For Microsoft, the deal fits into a broader effort to reconcile ambitious climate targets with the operational reality of AI growth. The company has pledged to become carbon-negative by 2030. Yet its own disclosures show the challenge is intensifying. In fiscal year 2024, Microsoft reported total greenhouse gas emissions of 15.5 million metric tons of carbon dioxide equivalent, up 23.4% from its 2020 baseline. The increase was driven largely by value-chain emissions tied to data centers, cloud infrastructure, and AI workloads. The company has not yet published its emissions figures for 2025.

That rise helps explain why Microsoft is aggressively contracting for carbon removal rather than relying solely on efficiency gains or renewable energy procurement. In fiscal year 2024 alone, it signed agreements covering about 22 million metric tons of carbon removals. Recent deals include backing AtmosClear’s project in Louisiana, which aims to remove 6.75 million metric tons of carbon dioxide over 15 years, and an agreement to buy 3.6 million removal credits from a biofuels facility owned by C2X.

The Varaha agreement adds a different dimension to that portfolio. While many carbon removal projects are based in the United States or Europe and rely on centralized industrial waste streams, India offers scale through agriculture. The country produces vast amounts of crop residue every year, and much of it is burned because farmers lack affordable alternatives. That makes India an attractive market for carbon removal developers, especially as companies search globally for projects that can physically pull carbon dioxide out of the atmosphere.

Yet operating at that scale introduces complexity that goes beyond building reactors. One of the biggest bottlenecks in carbon removal markets is not technology, but execution. Credits are only issued after projects meet strict measurement, reporting, and verification standards. Ensuring consistent feedstock supply, tracking material flows, and documenting outcomes becomes far harder when operations involve tens of thousands of farmers rather than a single industrial site.

Varaha’s co-founder and CEO, Madhur Jain, said Microsoft’s digital monitoring requirements forced the company to develop bespoke systems in-house. He noted that working with dispersed smallholders makes logistics and verification significantly more demanding than biochar projects in wealthier markets.

“More than 30% of our team has worked in agriculture,” Jain said, pointing to that experience as critical in designing systems that function on farms rather than only on paper.

The first reactor under the Microsoft deal will be located next to Varaha’s 52-acre cotton research farm in Maharashtra. There, the company already tests how biochar performs under real farming conditions, including its impact on soil health and yields. From that base, Varaha plans to scale to 18 reactors nationwide, with a total projected carbon removal volume exceeding 2 million tons over the project’s lifetime.

The company’s recent growth trajectory suggests it is racing to meet rising demand from corporate buyers. In 2025, Varaha processed about 240,000 tons of biomass, producing roughly 55,000 to 56,000 tons of biochar and generating around 115,000 carbon removal credits. A year earlier, annual credit generation stood closer to 15,000 to 18,000. Jain said the company aims to at least double its biomass throughput in 2026 to around 500,000 tons, with close to 250,000 tons of carbon sequestered.

Beyond biochar, Varaha operates 20 projects across India, Nepal, and Bangladesh, covering regenerative agriculture, agroforestry, and enhanced rock weathering. Fourteen of those projects are in advanced stages, with another six earlier in development. Together, they involve around 150,000 farmers and have the potential to sequester about 1 billion tons of carbon dioxide over lifetimes ranging from 15 to 40 years, according to the company.

For Microsoft, the strategic value of such projects lies in durability. Biochar is considered a long-lived form of carbon storage, which is increasingly favored by corporate buyers seeking higher-quality credits.

“This offtake agreement broadens the diversity of Microsoft’s carbon removal portfolio with Varaha’s biochar project design that is both scalable and durable,” said Phil Goodman, Microsoft’s carbon dioxide removal program director.

Even so, the numbers underline a structural tension. The volumes involved in individual carbon removal projects remain small relative to Microsoft’s overall emissions footprint, particularly as AI-driven demand continues to rise. Carbon removal is becoming less of a supplementary measure and more of a core pillar of corporate climate strategies, yet it is still playing catch-up with the pace of emissions growth.

Google is following a similar path. In January 2025, the company agreed to buy 100,000 tons of carbon removal credits from Varaha, marking its largest biochar deal to date. The parallel moves by two of the world’s largest technology firms show how competition in AI is spilling into competition for scarce, high-quality carbon removal capacity.

For India, the implications extend beyond corporate balance sheets. If projects like Varaha’s scale successfully, they could help reduce open-field burning, improve soil health, and create new revenue streams for farmers. Biochar application has the potential to lower dependence on chemical fertilizers, a significant cost for smallholders, while also addressing air pollution linked to residue burning.

Still, the broader question remains unresolved. Carbon removal projects take years to build, operate, and verify, while AI infrastructure is expanding at a much faster pace.