AI, the promise, the reality and the bursting bubble

I have been doing my own research into AI and data centres by talking to local experts and environmental scientists. I have also been getting useful information from other credible international sources.

Most people opposing the Southland data centre project are concerned about what is happening in the United States, where AI companies build these massive centres in small rural communities who do not have the ability to push back. The demands on energy, and water have an immediate impact and the lack of environmental controls also mean noise and light pollution too. From what I have gleaned the environmental impact of the Makarewa data centre won't have some of the extreme impacts of those in the US, but are still concerning without tight regulations and strong oversight. I am especially concerned about the three adjacent wetlands that could be impacted.

The biggest issue will be the electricity supply. When NZ is already struggling to have consistent supplies this one centre will be using 6% of our electricity production in two years. That is the same as what Hamilton, our 4th largest city uses with a population of 200,000.

When you look at the resources that our country will contribute to this centre and the benefits, they just don't stack up. You can look it up for yourselves, but New Zealand is actually known as the easiest country in the world for doing business in. Our Government welcomes just about any large corporate with lots of money and we are rapidly seeing our resources, production and profits head off shore. 50% of our forest cutting rights, 40% of our wine production and 20% of our dairy factories are overseas owned. We welcome mining companies but only charge around 1% in royalties on their profits and often have to clean up their mess afterwards. Our country has never been wealthier, our total wealth increases by billions each year but only a wealthy few and large corporates are benefitting.

This data centre is being built for private interests and will service global tech industries and minimal value will be returned for us. We seem to have stopped building stuff for ourselves under our own terms. We are told that we can't afford to do it ourselves and yet the government is prepared to build one motorway north of Auckland for $22 billion and provide $9 billion more funding to our military. Surely we could spend $1 billion on a smaller data centre to support our own highly profitable tech industry. It is just about choices and reclaiming sovereignty over our own important infrastructure. We already pay too much for our electricity, food and housing because too much is being sucked out of those sectors for private profit.

Then there is the wider impacts of the AI industry itself. According to the research of English academics (I will provide a link in the comments) governments and businesses are planning to layoff thousands of workers and replace them with AI. The economic impacts will be massive when we have skilled workers becoming unemployed or forced to work in lower paid jobs. The drop in local spending and lower tax income will impact on core services like health and education and further impact on our small businesses. This government has already decided to replace 9,000 state sector employees with AI, what job options will any of those have when even the private sector won't need their skills? It is also being revealed that AI may not be the cost saving that many thought.

It does seem that our government is blinded by the possibilities of AI and has no appreciation of the realities of the industry. It is actually based on a huge Ponzi like scheme where this is massive upfront investment but little likelihood of a profit. The industry has spent around $2.4 trillion on the data centres and infrastructure but is only making about $50 billion a year in income.

GPUs are the electronic processing units that manage the data in each data centre, but because the demands and technological advances are increasing so fast, they have to be replaced every 3 years. The constant retrofitting of data centres is a massive consumer of resources and money. The unit economics, or return on investment is shockingly bad. Most AI companies will most likely crash or liquidate, owing billions to their creditors. Who knows whether Datagrid will survive, even in the short term. It could be similar to the Great Financial Crisis in 2008 when taxpayer money was used to bail out banks. The bankers kept their corrupt earnings because they were seen as 'too big to fail' and the rest of us got poorer.
The Green Party hold the following positions regarding AI data centres:

Grid Strain and Power Prices: We are wary of the immense energy requirements of AI data centres—such as the massive Datagrid facility planned for Makarewa in Southland, which is projected to consume up to 6% of the country's total power supply. This massive demand could complicate the country's transition to a fully decarbonized national grid and hike up power prices for ordinary households.

Resource Management: We would hold data centres to account for their intensive resource consumption, including water usage, noise pollution, and land footprints, particularly in the regions they are built.

Corporate Accountability: The Greens advocate for digital sovereignty. We question why New Zealand should absorb the environmental and infrastructure costs of massive projects that mostly serve foreign corporate interests rather than enhancing local public and community services.

Australia is ahead of New Zealand on this issue and a number of organisations (NGOs and government entities) combined to minimise public harm. They came up with the following principles that should be expectations from any new AI data centre:

• Be powered by 100% additional renewable energy
• Strengthen grid stability
• Be appropriately sited to minimise impacts on nature and land use
• Minimise emissions and maximise efficiency and circularity
• Use water resources responsibly
• Operate with transparency
• Commit to earning and delivering ongoing social licence
• Support the training and up skilling of the workforce