As the technology world grows by the day, it’s relying more and more on the electrical industry to power it. Peter Vandenheuvel looks at what opportunities may be created by artificial intelligence for the electrical industry.

In early 2024, Elon Musk told the Bosch Connected World Conference that artificial intelligence (AI) will run out of electricity by 2025. It’s not 2025.

Whether true or not, several experts in the IT industry have mentioned the under-preparedness for AI from a power point of view. And where there are power problems, there are electrotechnology opportunities.

Electrical supplies at risk

An overstatement? Maybe, but when you consider the combined effects of moving to renewable power supplies with the introduction of electric vehicles (EV) with their electrical charging needs – all whilst at the same time phasing out of most (if not all) traditional fossil fuel generation – it will no doubt place electrical distribution infrastructure under much additional stress.

Even more so when also remembering that the electrical distribution systems are aging, not always maintained in their peak condition and they were never designed for the now changing distribution pattern.  It is moving away from the fewer large base load power station-focused feed-ins to the many more widespread feed-in supplies from rooftop solar, solar farms, wind farms and hydro arrangements.

The electrical network stress

There are multiple causes why the distribution systems are under stress. These include the less-than-optimal output and reliability of the renewables generation, both due to its limited ability to act as a base load capacity and its reliance on the fickleness of sun and wind.

It’s also the premature closing of fossil-fuelled base load power stations, as well as the continued ageing of the already overstretched electrical infrastructure.

Now add the already automatically increasing load demand brought about by the now already known other new technologies in the pipeline, such as:

• Take up of EVs, especially self-drives
• Focus on making factories eco-friendly
• Extra data processing for facial recognition
• Already under construction data centres
• Continuing growth in crypto farming
• Increased general data capturing
• Continued growth of online gaming
• Growing popularity of cloud data storage
• Moving from gas to electricity for heating

All these causes impact the electrical load directly, indirectly and/or with any further demands could create a perfect electrical overload storm.

Then along comes AI

There are great concerns the biggest AI growth effect will also add enormously to the already borderline-overstressed electricity generation and distribution systems.

Every time we add another substantial data user, we add further to the loading of the internet and the data centres that hold it up.

In Virginia, some experts say it, despite AI only being in its early stages: “Needs the equivalent of several large nuclear power plants to serve all its new data centres.”

It also indicated that at the same time, power companies are pushing commercial customers to go to extraordinary lengths to lock down energy sources, such as building their own power plants.

So, what happens when we really start to crank up the need for more AI memory? Well, if we are expecting all the tasks we do now to be offloaded and done by AI rather than ourselves, AI will need at least as much ‘intelligence’ for that task as we ourselves have now.

AI is hungry

AI is hungry for gigabytes, and that’s not insignificant as it is extremely data storage hungry. It will need all the gigabytes, terabytes and petabytes and the data centres to ‘store it all’ that we can throw at it.

According to Elon: “The artificial intelligence computing coming online appears to be increasing by a factor of ten every six months. That cannot continue at such a high rate forever, or it’ll exceed the mass of the universe, but I’ve never seen anything like it.”

Even though it may not ever be as drastic as mooted, there will be a huge increase in electricity demand, so here’s hoping its growth will slow down enough so we can keep up.

How many gigabytes?

Who knows? A quick search online tells us the average adult human brain has the ability to store the equivalent of 2.5 million gigabytes  – or 2.5 petabytes – of digital memory. That’s something to the order of 300 years of TV programs.

To try to put that into perspective, the same source advises the average desktop computer in our homes have a maximum of between 256GBs to 1,000GBs. So, if we try to compare that with a 256GB desktop, it would need 9,757 desktops to have the same memory storage as one human.

It demonstrates that even a few simple AI arrangements can add up to both enormous amounts of data storage and very much higher data traffic when in use. It is expected the AI-purpose processors and other infrastructure will require significantly more electrical power than a traditional data centre application.

The obvious solution

It’s simple, we need both more electricity and more data storage.

First – the extra power. This will require more power generation and equal growth in distribution infrastructure capability to cater for what will be the immensely increasing demand for electricity.

This will involve both new works and works to shore up the existing transmission and distribution networks as it will be a massive task to deliver all the electricity required to give life to the AI as well as equipment for all the other new competing technologies listed earlier.

That will need many more, and larger, solar and wind farms as well as hydro generation solutions, especially if the currently put forward green hydrogen production for electricity generation development hits too many hurdles and proves not viable. It may even mean needing to reconsider the current pushback on nuclear power generation.

Then – the extra data. This will involve not only the need for more new data storage facilities for both the storage and the huge increase in data transmission traffic. It will also require both new works and the upgrading of existing data storage and data transmission infrastructure to let the AI developers and those other technologies competing ‘do their thing’. It may even require innovative new solutions

Well, what’s in it for us?

This is great news for all of us involved across almost the entire electrotechnology business spectrum. These opportunities are not limited to any region, city or state as ongoing works will be required across all states and territories. So much of this will be happening on your own doorstep, no matter where you are.

It means more innovation, more new products, more engineering, more construction, more installation, more servicing, more manufacturing, more jobs and more opportunities for electricians to progress or start up their own businesses.

For the Electrotechnology industry, the development of AI could itself provide the industry with a bigger, longer lasting and more all-inclusive opportunity than any of the other initiatives now underway.

Opportunities may eclipse anything seen previously in the fields of power generation, solar farms, wind and hydro turbines, data centre construction and data transmission.

Your AI opportunities

There are six distinct streams of opportunity. For AI to be developed and realise its potential it will need:

1. Adding to the electricity supply capacity: The works involved in adding to existing generation equipment to providing new sources, which can be anything from adding more solar panels to any existing installations existing now to involvement in or undertaking new power stations or solar and wind farms and hydro installations. This also includes the installation of large high-power storage battery systems.
   Power generation could be more of the same – solar, wind and hydro – or with new technologies – green hydrogen and nuclear – to bolster the power supplies and avoid blackouts and brownouts. It may also require the strengthening of strategic intrastate HV supply links.
   An increased focus on additional rooftop solar for the late starting home and business owners as well as the life extension, upgrading and renewal of the already installed solar systems, especially for those installs connected at the early introduction and now reaching their use-by dates.
2. Adding to the electrical distribution capacity: The works involved in upgrading and extending the current overhead and underground infrastructure, which can be anything from upsizing the feed-in to a home or upsizing cables in a street to installing new or changing-out existing transformers in the street to upgrading or building entire new substations.
   Upgrading and extending local power distribution systems, underground and overhead, with large high and low-voltage cables, all manner of transformers, switchboards, more poles and wires, distribution plant, more substations and or upgrading of existing substations, etc.
3. Adding more data storage facilities: The works involved in upgrading or building new data centres to satisfy the AI hunger for data capacity, which can be anything from minor works in small data centres or customer facilities to upgrading the electrical installations to existing data centres to building and wiring entire new centres.
   There will be many more new data centres and life extensions, upgrades or expansions of existing ones. All this requires new or upgraded large standby generators, batteries and uninterruptable power supplies, as well as complex data cabling and fibre installations and massive switchboards and mega-multiple data panels, air conditioning, etc.
4. Adding more data capacity: The work involved in expanding the data capacity in existing facilities by adding more data panels and associated equipment and the additional power and data installation.
   This can also include upgrading existing air conditioning as the new AI processors and peripherals may generate more heat.
5. Upgrading fibre to the premises: The work involved in upgrading the network connections to fibre, which can be anything from replacing the copper cabling to fibre for single users to undertaking work for organisations such as NBN or Telstra.
   Additional upgrades and new construction of data transmission systems, be they fibre in-ground or by wireless transmission towers, all also need power supplies, switchboards, control panels, etc.
6. Power and or data installations for AI-developed equipment: The works involved in installing electrics and data to equipment developed for – or upgraded to – use and operate the AI as and where it requires robotics or similar to deliver its functions.

Who can benefit?

Almost all of those persons and enterprises involved in the electrotechnology industries include – but are not limited to:

• Electrical contractors and electricians involved in installation, upgrading, upscaling and servicing of electricity generation, electrical utility infrastructure and/or data centres
• HVAC contractors for upgrading existing installation and installation in the new data centres
• Data centre equipment, cabling contractors and data panel suppliers
• Manufacturers including switchboard builders, power transformer manufacturers, HV and LV cable makers, large UPS and electrical generation suppliers and installers, HVAC units, solar panels and wind turbines
• Manufacturers, suppliers and installers of medium and large high-power storage battery systems, as well as for roof-top solar installations including on the data centres
• Electrical wholesalers and other suppliers for other installation materials such as cabling, cable racks and so on
• Engineers and consultants who are specialists in generation, distribution, large storage battery and construction electrical and data installations

Take your pick

If you are interested in – or looking to take part in – any of the wide-ranging opportunities that will be on offer, think of the skills you already have and those you want to develop and at what level you want to partake.

This can range from subcontracting or supplying goods and services to those already heavily involved in the industry.

And it will keep on giving

Unlike the one-off day-to-day projects or initiatives, many in the industry rely on AI development is not only likely to keep accelerating at the current rate or greater. But given the nature of it, AI will continue to grow for a long time. So, for those who want to be involved and put in the effort to do so, it can be the gift that keeps on giving for many years, so please don’t reject this opportunity without giving it some serious thought.

When does it all start?

It already has, but there is still time. Once it does start in earnest, it will take off rapidly and then will be too late.

It will take some effort due to the highly specialised nature of the generation, utility and data centre work requirements. Waiting to see what happens before deciding if you want to be a part of will leave you way behind at the back of the pack – and it will be most unlikely that you can catch up.

Getting a ticket to ride

Working in the power generation, electrical distribution and/or data centre fields are highly specialised and may require experience, some forms of accreditation or other references that will be requested by the companies you apply to.

Unless you are known in those fields, you are unlikely to get any meaningful work without some effort. So, find out what the expectations of your target customer are and get this sorted out now.

If you are looking at involvement with AI activity supporting companies and utilities, the time to start lifting your company, personal profile and recognition of your skills is now, especially if doing the work is borderline your current comfort zone.

Getting your house in order

As alluded to above, to undertake work of this nature in these specialised areas you must at a minimum have robust, recognised and regularly audited third-party accredited quality and WHS systems in place.

These should also include procedures and processes that incorporate detailed checklists applicable to the types of work to be encountered, as well as comprehensive risk analysis –A strictly adhered to ‘no live work policy’ should also be embedded in your systems.

Closing any identified gaps in your skills and having a policy on training that includes the skills needed in your chosen preferred customer’s areas is also highly recommended.

Finding customers of interest

It’s obvious that primarily – unless you plan to get involved by actually doing your own AI development – you will not be looking for companies actually doing AI development except as in the six mentioned above.

As an electrotechnology company however, you will likely be looking to work for (or with) those major general contractor companies already providing one or more infrastructure support activities to the electricity generation, power distribution, data centre construction etc associated activities that see a benefit in subcontracting those works.

However, that’s not to say you may not also find and work for AI developers as in the six above that have a need for electrotechnology works on the AI hardware requirement in a more direct way, but they might be more challenging to identify.

Once you have found the potential customers, it is suggested you build relationships to a level where the customer’s decision-makers have confidence in you and your ability to deliver to their expectations as soon as practicable.

Please remember that AI opportunities are vast, wide-ranging and there for a long time, so make sure you get your slice of the AI pie.