Some Are Starting To Re-think Their ‘Net Zero’ Plans

Written by Bret Swanson on September 21, 2024. Posted in Current News

Last week, Mario Draghi, the former president of the European Central Bank, sounded the economic alarm. In a 400-page white paper, Draghi warned an uncompetitive European economy faces “an existential challenge” of flagging dynamism and slow productivity growth

The start-up ecosystem is paltry, and energy prices are far too high. Europe will have to invest an additional €800 billion per year in infrastructure, Draghi urges.

But where will that money come from in a slow-growth economy, and how can Europe build more without consuming more energy and jeopardizing its “decarbonization” targets?

Draghi admits it’s a genuine dilemma, one that most European leaders have put off acknowledging for far too long.

Larry Fink has also been mugged by reality.

For years, Mr. Fink, CEO of BlackRock, the world’s largest financial firm and shareholder, led the charge for environmental, social, and governance investing, or ESG. In increasingly urgent annual letters beginning in the late twenty-teens, he called for “a fundamental reshaping of finance.”

He suggested a shift to “stakeholder” instead of shareholder investing. He demanded companies in every industry push power sources such as windmills, and punish hydrocarbons, to meet ‘net zero‘ carbon dioxide emission targets.

This spring, however, Mr. Fink reversed course. “The world is going to be short power,” he acknowledged. You can’t run the artificial intelligence (A.I.) revolution on “intermittent” wind and solar energy, he quickly followed up.

For A.I., you need abundant, readily “dispatchable” power, he insisted. Welcome to the real world, Mr. Fink.

The reversal is good news for the energy industry and broader economy. But it also is a reminder of the dangers of corporatism – the accelerated trend in which large firms partner with government to achieve policies the public rejects.

If Mr. Fink hadn’t woken up to the energy realities of A.I. computation, he may, from his perch atop the global investment world, have pushed ‘net zero’ energy poverty for many years to come.

Europe, which embraced ‘net zero’-like policies over the last two decades, is today suffering the results, including slow wage growth and deindustrialization.

Europe leads the world in ‘renewable’ electricity generation but often pays two to three times as much, or more, per kilowatt hour as the U.S. (The same is true for California, by the way. See chart below.)

By one measure, Eurozone productivity over the last quarter century grew just one-third as fast as the U.S. The result, Draghi concludes, is that “real disposable [per capita] income has grown almost twice as much in the US as in the EU since 2000.”

The divergence is especially alarming because U.S. productivity growth over this period was itself sluggish. Fortunately, beginning around the time of the Global Financial Crisis, America embraced its massive shale hydrocarbon resources and hydraulic fracturing technologies.

The resulting energy explosion helped offset numerous policy mistakes.

These energy resources will be central to the A.I. story. And the success of A.I. will itself largely determine the pace of innovation and wealth creation across every economic sector over coming decades.

Draghi does not nearly reverse course on Europe’s climate orthodoxy, but he does subtly acknowledge a key dilemma. He writes that offloading energy intensive tasks to China helps Europe ‘decarbonize’ and meet its ‘net zero’ targets.

But this very off-loading of manufacturing and technology deprives Europe of important innovation, and thus productivity, pipelines.

Generating artificial images, machine-written essays, software code, computer vision algorithms, or protein folding simulations requires massive amounts of electricity. To achieve these miracles, A.I. companies must first train “models” with exabytes of data culled from all historical Internet content.

In 2024, a leading edge A.I. training cluster consisting of 100,000 chips might consume 100 megawatts of electricity, or enough to power 100,000 homes. By 2026, one A.I. training warehouse might contain one million chips and consume one gigawatt, or the capacity of a large nuclear reactor.

Training A.I. models, however, is only the beginning. The second half of the equation is “inference” – asking the models to generate answers. Delivering the services you and I want – reading, digesting, answering, and learning from our requests – will require yet more computation and power.

The former OpenAI researcher Leopold Aschenbrenner predicts that by 2030, the four large A.I. companies will each spend $1 trillion to build training clusters with 100 million chips consuming 100 gigawatts, or 20 percent of current U.S. electricity production.

That’s an insane prediction, you reasonably reply. Surely, such a system would collapse under its own weight. Won’t A.I. overexhuberance go bust long before then? Yes, a popped bubble is not unlikely, and Big Tech stock prices have already moderated from their stratospheric summer highs.

But remember the technology-telecom crash of 2000-01, when we rapidly built millions of miles of Internet links without the content and commerce to  immediately fill them up?

Although hundreds of dot-coms and dozens of major fiber-optic networks and broadband companies went bankrupt in the short-term, those losses today look mild compared to the size and transformational impact of today’s  Internet economy.

Likewise, even if today’s A.I. hype leads to a mid-term crash, the data center and energy infrastructure we build today will determine the success of the economy for decades to come.

Refining information into intelligence is the central force of all economic growth. It was true of the printing press, the telegraph, the computer, and now A.I. But information refineries require energy refineries – and power generation and transmission and resilience, all at reasonable cost.

Bits need atoms, and atoms aren’t free. The availability and cost of energy will in substantial part predict the advance of A.I.

‘Net zero’ energy policies, enforced by undemocratic ESG muscle, will scuttle this future. Europe already shows a preview.

Fink may have seen the light, which Draghi is finally, if only partially, acknowledging, possibly too late.

See more here substack.com

Please Donate Below To Support Our Ongoing Work To Defend The Scientific Method

PRINCIPIA SCIENTIFIC INTERNATIONAL, legally registered in the UK as a company incorporated for charitable purposes. Head Office: 27 Old Gloucester Street, London WC1N 3AX. 

Trackback from your site.