• This 17th century proverb sums up big tech’s approach: “Fine words butter no parsnips”
• AI: The electric elephant in the datacentre
• Once quick and responsive, big tech is now slow, bureaucratic and a global problem
• Political changes in Europe, the UK and the US could presage meaningful control over a sector that has evaded scrutiny for too long

“To begin at the beginning”, as Dylan Thomas had it not that long ago, the big tech behemoths that now bestride the globe were start-up, underdog disruptors of a status quo that had long been established, and seemingly impregnably fortified, by the likes of Microsoft, Apple and their ilk. Facebook (now the ‘wanna-be’ all-encompassing Meta), Alphabet (Google), Amazon, etc. created new business models with the intent of taking market share from ‘establishment’ corporations by gaining access to the personal data willingly (and permanently) donated by millions upon millions of their users in return for access to ostensibly ‘free’ services.

The data gathered, manipulated and reused was then weaponised to blunt and stunt competition as immensely rich, de facto monopolies found themselves able to acquire and then exploit or destroy any upcoming next-generation companies deemed either to be a threat to their hegemony or possessors of technologies that could be of use in furthering big tech’s expansion into other areas or sectors.

By this means, continuing double-digit growth has been achieved and with that has come massive profitability and big tech’s member companies’ perceived requirement to maintain their places in the hierarchy by any means at their disposal.

Meanwhile, throughout the inexorable rise of these companies, their regulation, via competition laws and national, regional and international legislation, has been sluggish, partial and remarkably ineffective – not least because, alongside the creation of wealth and profit, the companies have grown so big they have also accrued to themselves immense market power way, way beyond what has been allowed in other industries, such as finance and insurance. And with that has come an iron-fisted political clout usually exercised via the velvet glove of pervasive lobbying power.

So far, efforts to curb the excesses of the big tech companies have been to little avail, but as AI spreads and the impending impact of the law of unintended consequences looms, calls to exert control over companies that, in many ways are already monopolies and are beginning to take on the attributes of oligopolies, are growing and regulators and lawmakers are at last moving to introduce meaningful control that could see the likes of Meta and Google broken up into constituent parts. Well aware of this, big tech is casting about to prove that it is not a threat and has humanity’s best wishes at its heart. One tactic is to burnish and reburnish its green and sustainability credentials.

Last week, journalists received a spate of press releases from Amazon, in which the company claimed it has already bought and secured enough clean electricity to meet all the energy demands of all its datacentres, headquarters and regional offices, and warehouses across the entire world – a full seven years ahead of its ‘sustainability target’.

The claim comes even as climate scientists have begun to concur that a planned longer-term route to net zero would be more important and effective than a quick dash from apparently equalising every tonne of emitted carbon dioxide with countervailing actions to mitigate such discharges and then going on to claim to be carbon neutral (and claim kudos accordingly). Interestingly, Google has admitted that its AI operations last year resulted in a 13% increase in its greenhouse gas (GHG) emissions and now no longer claims to be a carbon neutral company. Given Amazon’s huge plans for AI, we also need to know what effect they will have on its emissions.

Quick, simple and cheap answers trump the slower, complex and more expensive solutions every time

The trouble with most extant corporate plans to reach net zero is that they are complicated and costly. They usually require participants to either completely balance out every tonne of carbon dioxide (or other GHG) emitted – and one quickly effected and relatively inexpensive way to negate emissions is to buy carbon credits to ‘offset’ them. In essence, this means paying some company or other to plant trees, clean up polluted industrial sites or lobby to get the use of agricultural pesticides and chemical fertilisers reduced or banned.

With the best will in the world, these solutions, which can never be more than partial, are very difficult to police or quantify and will take many years to lower emissions even by a few percentage points. Now, the compass is swinging round to the point where the focus will be on requiring big tech itself to remove its own emissions at source, and that’s going to cost them a great deal of money.

Amazon says it hit its clean-electricity target impressively early because, it argues, it is lowering emissions, improving energy efficiency, purchasing more guaranteed carbon-free power, building renewables projects at its facilities and supporting similar projects around the world. This has been achieved by “purchasing additional environmental attributes (such as renewable energy credits) to signal our support for renewable energy in the grids where we operate, in line with the expected generation of the projects we have contracted.”

However, buying renewable power generation that happens at some undetermined time or other in some far-flung location is not the same as sourcing and using the amounts Amazon does in its continuing and continuous operations in datacentres and “fulfilment centres” (warehouses) at specific times.

Amazon is well known for its antipathy to trades unions, but its ‘Employees for Climate Justice’ is a group of more than 8,700 Amazon workers that has had some success in forcing the company to pledge to take action to mitigate climate change.

Back in December 2018, the nascent Amazon activist group was able to file a shareholder resolution tasking the company to create “a comprehensive, company-wide plan to reduce its fossil fuel dependency.” In response, Amazon did very little until the staff announced that they would down tools and the publicity surrounding the planned walkout was sufficient for the company to announce that it intended to achieve net-zero carbon emissions by 2040. That’s still 16 years away and the ‘Employees for Climate Justice’ group is not impressed with the company’s progress to date.

Notwithstanding Amazon’s considerable investments in renewables, including plans to make its entire enormous fleet of delivery vehicles electrically powered, the ‘Employees for Climate Justice’ calculated that 78% of the energy Amazon consumes in the US still comes from “non-renewable sources”. It adds that the company uses “creative accounting” to substantiate claims that it has hit its clean-electricity target.

The fact of the matter is that the planned proliferation of AI-powered datacentres is actually increasing the demand for fossil fuels. And there’s the rub. AI, on which much of the future development and success of big tech depends, will rely on a massive consumption of energy. For its part, Google has been quick off the mark to claim that it will use AI to find ways to lower emissions, but the fact remains that AI-dependent datacentres will require the generation of evermore power into the foreseeable future.

As TelecomTV reported recently, an indication of the size of the problem can be gained from research by Polar Capital, the London, England-headquartered specialist investment management company that runs the Smart Energy Fund. It says the installed datacentre capacity for AI training will multiply from the 5GW or so consumed worldwide last year to more than 40GW by 2027. That 8-times multiple is more than the current annual power consumption of the entire UK. Meanwhile, research house Gartner estimates that generative AI on its own could consume 8% of all electricity generation anywhere – and up to 3.5% of the entire world’s electricity supply – by 2030.

Good job it’s getting windier.

- Martyn Warwick, Editor in Chief, TelecomTV