The key problem that policymakers,
investors and, indeed, academic economists face is that with total global debt having reached
three times global GDP at the time, and with much of that debt associated with older industries that are capital, raw materials, facilities and labour ‘intense’, these new industries may prove to be so ‘disruptive’ as to leave the older capital intense (i.e. debt-laden) industries staring down the barrel of a gun with very profound consequences for price formation, the financial sector and the global economy.
Ten years later, the evolving technological revolution finds it focal points in the AI and energy transformation booms, as well as facing a sharp shift in financing, energy and raw materials needs from the then minimalism to the current explosive CapEx trend. This threatens to tip the balance from deflation to inflation, in no small part exacerbated by the prior period of underinvestment in upstream supply chains and infrastructure. The current need for massive investment in scientific research, labour skills training, production and processing facilities, sourcing and extraction of raw materials including ensuring secure and stable supply chains, expansion of facilities, distribution and logistical infrastructure, and above all sustainable and affordable energy supplies is all too obvious and imperative. But the sharp pendulum swing from ‘under-‘ to likely ‘over- investment’ cannot ignore a lack of visibility on short, medium and long-term demand, the fact that technology will be rapidly superseded, and that the herd like instincts of the financial world are not only fickle, but often poorly informed and frequently misdirected. Ironically, for the first time in 20 years, the November 2025 BofA Securities survey highlighted investor concerns that companies are ‘overinvesting’.
Two elements of the current AI related investment are of particular concern. For many a year the tech behemoths had strong free cash flow, which helped to justify lofty valuations relative to many other sectors. But the surge in investment spending, accompanied by jumbo bond issuance (as per the recent $30 Bln issuance from Meta and $23 Bln from Alphabet) now means that investment spending is outpacing revenue growth by a factor of 5:1, and by extension free cash flow is shrinking rapidly, especially when one considers ongoing share buyback programmes. But rather more worrying are changes to accounting practices, which flatter EPS, but also fly in the face of reason.
Specifically the practice of extending depreciation periods for assets, when it is obvious that the pace of technological evolution has picked up, and therefore technology assets should therefore be depreciated at a faster rather than a slower pace. When one then considers that a ballpark estimate of how many technology start-ups fail is around 85%, and then thinks about the implications for how much of this current investment bonanza may have to be written off, there is good reason to be cautious.
AI technology is also energy hungry, thus requiring any companies involved in this to also invest heavily in obtaining reliable, sustainable and secure supplies, which has understandably seen many turn to nuclear. Anecdotal evidence suggests that a lot of this at prices that are above regular PPA (Power Purchase Agreement) rates, effectively aiming to guarantee long-term supply, and facilitate providers investments. But this will likely crowd out some investment in ‘regular’ power generation and distribution, and add to inflationary pressures, given the long run underinvestment in grid infrastructure, and a likely parabolic rise in electricity demand in the context of the energy transformation. Now throw in the consideration that critical minerals supply is a major issue, above all because the processing of these minerals is not only controlled largely by China but also requires hefty investment in processes which are currently environmentally unfriendly. It should be also noted that while there has been a sharp increase in investment in recycling industrial metals, particularly copper, much of this recycled output does not meet the quality (in this case purity) requirements for manufacture of many key elements for relevant parts and equipment. Per se there is no room in all of this for wishful seeing or wilful blindness. That is not to say that technology will not evolve to meet these needs, but probably not at a pace that meets demand. The likelihood is that corners will likely be cut to meet enormous immediate demand, which may result in reliability and safety issues, and in tum demand regulatory interventions, which will probably lag behind the technological development curve. The simple point is that the need for systemic thinking has never been higher, and that linear solutions-oriented thinking may prove to be a very substantial risk.
Marc Ostwald E:
marc.ostwald@admisi.com T: +44(0) 20 7716 8534
AI’S ENERGY DEMANDS AND RELIANCE ON NUCLEAR POWER RISK CROWDING OUT TRADITIONAL POWER INVESTMENTS AND FUELING INFLATION, WHILE CRITICAL MINERAL SUPPLY REMAINS A MAJOR ENVIRONMENTAL AND
GEOPOLITICAL CHALLENGE.
19 | ADMISI - The Ghost In The Machine | Q4 Edition 2025
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32
