Is the Firmus fiasco a crack in the AI boom?

Australia may be good at innovation and invention (think wi-fi and over-the-horizon radar, along with a host of medical breakthroughs and advances), but as a nation we don't have a great track record in attracting investment to fund those brilliant ventures.

Traditionally they've headed offshore.

Amidst the wreckage of last week's failed mega capital-raising by data centre provider Firmus Technologies was a creeping sense that, once again, we had somehow fallen short.

Perhaps there's a less philosophical and simpler explanation for the embarrassing failure to launch.

Big investors, including both domestic super funds and global hedge funds, baulked at the price and deemed the entire venture too risky.

Why?

This is a company still in its infancy that churns through capital and racks up losses, big on promises and short on detail, demanding to be valued at $44 billion.

Or perhaps they were thrown by the sudden decision to alter course in the days preceding the sales process, abandoning a multi-billion-dollar joint venture with an Australian data centre group to focus instead on Asian expansion.

And then there was the management team. While there has been an endless stream of words churned out about the criminal record of one of the founders, a far more important issue loomed large.

Both Oliver Curtis and his partner Tim Rosenfield have barely any history in technology, precious little experience in data centres and are novices when it comes to the daunting task of running a multinational public company.

The biggest selling point for the float was that some of the world's biggest investors, including American giant Blackstone, and industry titans like Nvidia, were on board, delivering credibility to the Australian-born Firmus.

That begs the question: was the Firmus float collapse a sign that the AI investment bubble is running short of fuel?

Data centre backlash will delay AI earnings boom

Maybe Elon Musk shares some responsibility.

Just 16 weeks ago, his SpaceX conglomerate — a mix of AI, robotics and ah, social media — soared into the history books with the world's biggest Initial Public Offering.

It raised an eye-watering $US85.7 billion, valuing the company at more than $US2 trillion on day one and catapulting Musk into rarefied territory as the world's first trillionaire.

Like Firmus it burns cash. Like Firmus it is fuelled by hype and vague promises about the future, including outlandish plans to build data centres in space.

As fanciful as Musk's ambitions seem he may have hit upon something important.

For across the developed world a huge backlash is growing for data centre construction. 

He should be able to avoid that problem on Mars, or even the Moon.

Across the United States, disgruntled residents are marshalling communities to halt the rapid roll-out of data centres, with one poll by Gallup finding that 71 per cent of Americans are opposed to the massive barns filled with high powered electronics.

To put that into perspective, only 53 per cent are opposed to nuclear power plants.

A similar trend is sweeping across Europe with more than 70 data centres rejected or restricted so far this year.

Meanwhile, here in Australia, the data centre revolution is being put the test in capital cities across the nation.

Everywhere, the complaints are the same.

Huge amounts of power and vast amounts of water are required to keep these massive sheds perennially humming.

You could dismiss it as NIMBYs (not in my backyard) selfishly grumbling about the future while they demand ever more sophisticated technology.

But the opposition places a unexpected roadblock in the way of the AI rollout.

That's something the tech bros, and their supporting army of investment bankers, never took into account as they hyped a boom that has been driving global investment markets now for four straight years.

The big tech corporations at the heart of the artificial intelligence boom have been driven to stratospheric valuations underpinned by perfect execution, that nothing could go wrong.

There's no contingency for delays and missteps, even temporary ones.

Boom or kaboom?

Wall Street is in the final throes of a four-year boom.

It's not unheard of. But it's rare.

Generally, such booms either start from a position of weakness, such as the eight-year winning streak to 2016 after the global financial crisis, or end in market crashes such as the 1987 crash and the dot-com collapse in 1999.

The pandemic may have rudely interrupted this boom as tech stocks were powering the pre-pandemic market to new records.

But the AI revolution has sent a handful of already seriously overvalued stocks to new heights, making it one of the most unevenly balanced stock booms in history.

Not everyone agrees that this is a bubble. Supporters point to the rapid growth in earnings from the technology giants.

What isn't in dispute is that the race for technological dominance has been undertaken at breakneck speed. But getting it to the customers may not be as easy.

While the technology is being developed unhindered in labs and basements within Silicon Valley and beyond, the task of delivering it to the masses involves winning over communities while navigating a host of local, state and national government rules that could sideline projects for years.

It is akin railways. You can build the world's best locomotive and carriages. But without the tracks you ain't going nowhere.

'An existential risk to humanity'

Has there ever been a boom like this, where the leaders continually warn of a potential calamity from their own endeavours?

Anthropic's Dario Amodei, in particular, has been on the parapet urging caution, even as his models plunder databases in other countries like the recent hack of our very own Medicare.

OpenAI's Sam Altman has been equally cautious about the technology's application but, unlike Amodei, he doesn't quite see the need to slow the pace of AI development.

Amodei and some of his colleagues embrace a philosophy called effective altruism, and harbour genuine concerns about the future.

But there is also a view that the warnings are partially motivated by Anthropic and OpenAI's need for new capital.

Unlike the mega-cap tech giants, both AI specialists can't easily tap their own shareholders, are forced to rely upon debt and private investment markets, and are now trying to line up stock market listings.

They also rely heavily upon the technology giants for investment funds, while concerned the very same giants will eventually just devour or overtake them.

SpaceX's performance since listing hasn't done them any favours.

It's been anything but consistent, initially rising, falling heavily and then recovering, sowing seeds of doubt in the minds of potential investors for future IPOs.

The Firmus fiasco has only heightened those fears.

Both OpenAI and Anthropic rely upon tech giants Microsoft, Amazon, Nvidia and others in what has become an increasingly circular investment pathway where investors supply product and buy services.

While each are feeding off the other as growth accelerates, there are genuine concerns that were raised a fortnight ago by the Reserve Bank of Australia about the potential for contagion if something goes wrong with just one major player.

So far, the rising cost of debt has done little to dent the AI boom. 

But cracks are beginning to emerge in the glittery AI façade.