These Aussie startups are rewriting the green AI standard


Welcome back to Neural Notes, a weekly column where I look at how AI is affecting Australia. In this edition: as global tech giants’ emissions keep rising, some Australian startups are setting the new ‘green AI’ standard.

We need to stop ignoring AI’s environmental effects

AI is transforming industries.

This is a line we have been hearing ad nauseum for almost three years. 

And it’s true, but it comes at a steep price for the climate.

Google’s own environmental disclosures reveal its magnitude. In 2024, its greenhouse gas emissions hit 11.5 million tonnes CO₂e. This was an 11% increase from 2023, which had already jumped by 13% year-on-year.

To look at it another way, Google’s emissions in 2024 were 51% higher than in 2019, “largely due to the explosive growth of AI workloads and expanded data center operations”. 

Even with efficiency breakthroughs and record clean energy contracts, both electricity and water usage keep rising.

Related Article Block Placeholder

Article ID: 316008

The rest of Big Tech is on a similar path. Meta’s emissions have doubled since 2019, surging to 14.1 million tonnes CO₂e by 2023, with Scope 3 emissions rising 65% in just two years.

Microsoft’s footprint grew 40% between 2020 and 2023, hitting 17.1 million tonnes. Amazon’s operational emissions have risen sharply as well, reporting nearly 69 million tonnes CO₂e in 2023, driven by cloud and AI expansion. 

As Microsoft itself put it: “The increased energy demand from AI workloads makes meeting our climate goals much more challenging”.

The shifting tone is increasingly hard to ignore. Big Tech’s boldest sustainability messaging — once front and centre — has grown muted as AI’s energy appetite outpaces efficiency and renewable rollout. 

Research shows generative AI queries use 10–33 times more energy than a standard search. This is likely to rise further due to AI summaries becoming the norm for Google searches. 

By 2023, data centres accounted for up to 1.5% of global electricity consumption, a figure likely to triple by 2030. 

With these pressures mounting, real transparency is getting harder to find. The most critical, hardest-to-measure layer, Scope 3 emissions, remains the least visible in public reports.

What do Scope 1, 2, and 3 actually mean?

To understand the real climate effect of AI, it’s important to break emissions down into three categories (known as “scopes”) used in Australia’s National Greenhouse and Energy Reporting (NGER) Scheme and worldwide:

Related Article Block Placeholder

Article ID: 278378

Scope 1 covers direct emissions from things a company owns or controls. This includes the likes of fuel burned in backup generators at data centres, company vehicles, or on-site leaks.

Scope 2 includes indirect emissions from purchased electricity or utilities. For example, the power used to keep servers, offices and energy-hungry AI workloads and cloud infrastructure running.

Scope 3 captures all other indirect emissions across a company’s value chain. For example, manufacturing and shipping servers and chips, business travel, product use, and infrastructure demolition.

For big tech, Scope 2 and Scope 3 are where most AI-driven climate consequences are now concentrated. 

In Australia, companies that meet significant emissions or size thresholds are legally required to report Scope 1 and Scope 2 emissions under the NGER Scheme. 

However, starting from 2025, new climate-related financial disclosure laws will require large Australian companies, including listed and unlisted firms, super funds, and investment schemes meeting the Corporations Act thresholds, to report Scope 3 emissions. 

Australian startups leading the way on green AI

As tech giants wrestle with rising emissions, some Australian startups are setting a different benchmark by deploying AI for real ecological and business benefits, and being open about their own digital footprint.

In finance and agriculture, WollemAI (formerly Wollemi) is focused on eliminating guesswork from climate accounting. 

Its platform uses machine learning to synthesise asset-level emissions data, enabling banks, insurers, and investors to move from ticking a box to real climate action.

In a Suncorp Bank pilot conducted by the startup, the tool helped reduce farm-level “financed emissions” calculations by up to 97%.

“Unless you’ve got the accurate measurements, you don’t know what you’re doing,” CEO Sam Sneddon told SmartCompany in 2023.

“The carbon accounting space is really noisy. But we haven’t seen anybody who’s been able to achieve what we’re doing specifically for land and agriculture. And that’s taking machine learning and science-based to establish highly granular emissions management.”

Sydney-based Xylo Systems, led by CEO and co-founder Camille Goldstone-Henry, is also at the forefront of this new green standard.

Related Article Block Placeholder

Article ID: 291391

Unlike conventional carbon accounting, Xylo’s AI-powered biodiversity intelligence platform helps companies go deeper. It measures the ecological footprint of a business, including the local flora and fauna, and provides data-driven steps on how to help regenerate those ecosystems.

And it’s done using AI. 

“We use biodiversity data to help businesses become more nature positive,” Goldstone-Henry told SmartCompany.

“Biodiversity is facing an absolute crisis. There’s a million species at risk of extinction and unchecked economic development is contributing to that. So now there’s a lot of pressure on businesses to understand and mitigate their impact on the environment. 

Goldstone-Henry said Xylo’s intelligent data platform could go beyond carbon and climate reporting to address the ecosystem and species a business may be affecting, both in the immediate vicinity of its locations and across its supply chain.

One tangible example is Xylo’s recent work with a multi-billion-dollar property company in Singapore.

Using Xylo’s platform, the business discovered properties with more green space not only saw measurable biodiversity benefits, but also experienced higher tenancy rates and greater yields. In other words, making green decisions had a positive effect on its bottom line.

What makes Xylo’s approach particularly interesting is its candour about the environmental costs of building digital solutions. 

“With Xylo being a for-purpose business, we have a very strong theory of change. And one of the big risks to [that] is the environmental impact of not just AI, but cloud computing in general,” Goldston-Hemry said.

“We rely on huge amounts of data, and that requires a lot of cloud computing. So even regardless of AI, we’re having some sort of environmental impact.”

One of the things Xylo does is keep its algorithms as efficient as possible, and opt to use the “most green” cloud providers possible.

Xylo also just released its own Nature Disclosures Report, which is unusually transparent for a young digital startup. 

“At Xylo Systems, our mission is clear: help companies make better decisions for nature. But to do that with integrity, we must also hold ourselves accountable,” the report reads. 

The report recognises Xylo’s largest environmental impacts sit upstream in cloud computing and hardware supply chains. Despite being a comparatively lean company, it still has an effect on natural resources.

Xylo is proactive about addressing its footprint by targeting net-zero emissions by 2030 across all scopes, extending device lifespans, tracking biodiversity around its headquarters, and developing new metrics so clients and partners can measure real nature-positive outcomes. 

In an AI sector often focused on efficiency and workloads, Xylo is showing real green AI leadership means not only helping clients do better, but also setting a new standard for itself in what sustainable tech should look like.

And as environmental reporting standards tighten, more businesses will need to pay closer attention to this playbook.


Source

Visited 1 times, 1 visit(s) today

Recommended For You

Avatar photo

About the Author: News Hound