A massive, international effort to research and develop carbon-free energy technology "pursued with the urgency of the Manhattan Project or the Apollo space programme" will be needed, if the world is to stabilize carbon-dioxide emissions, at the same time as maintaining the global economy into the next century.

In a report in Nature, Martin Hoffert of New York University, New York and colleagues look at the world's projected requirement for carbon-free energy, as nations seek to put a brake on carbon-dioxide emissions – implicated in global warming – while maintaining economic well-being.

The researchers take, as their yardstick, a goal of stabilizing the concentration of carbon dioxide in the atmosphere at around 550 parts per million by volume (ppmv) – about twice the concentration that it was before the industrial revolution, when fossil fuels started to be consumed on a large scale, and about 50% more than the current atmospheric concentration. Even taking into account the significant increases in energy efficiency (about one percent per year) seen in recent years, stabilizing the atmosphere at 550 ppmv will require about 15 trillion watts (terawatts, or TW) of carbon-free power by the year 2050. To put this in perspective, the world currently consumes about 13.5 TW of which no more than 3.4 TW are 'carbon-free'. 'Carbon-free' power means any source of power in which fossil fuels (coal, oil and gas) are not consumed, releasing carbon dioxide into the atmosphere. Nuclear power, wind power and solar power are all classed as carbon-free.

The researchers' estimates are based on a formula called the Kaya Identity, in which four variables are multiplied together: the number of people; the per-capita gross domestic product; the amount of carbon dioxide released per unit of energy (the 'carbon intensity'), and the energy consumed per unit of economic output (the 'energy intensity').

Only the last two are considered in the Nature report, leaving policy-makers with three choices – reduce energy intensity, reduce carbon intensity, or both, while simultaneously doing as little as possible to harm the global economy.

The news is not all bad: over the past 150 years, the tendency has been to reduce energy intensity (that is, to improve energy efficiency) and carbon intensity (by, for example, switching from more to less 'carbon-intense' fuels, such as moving from coal to natural gas. But current improvements will not be enough to meet an increasing need for carbon-free power to maintain the economy, given strictures on carbon-dioxide emission and an increasing pressure of population.

For example, say the researchers, the evolving global energy mix, based on declining costs of nuclear and carbon-free energy relative to fossil fuels, has the carbon intensity dropping to that of natural gas by 2030. But this can only be achieved by a massive introduction of carbon-free power, of the order of 10 TW by 2050. This holds even with significant efforts to improve energy efficiency.

The researchers imply nothing less than a wholesale movement towards carbon-free energy technology to produce more power, carbon-free, than the world currently consumes. This requires the kind of massive investment in R&D that only national governments – or international bodies – are able to supply.

The picture here is gloomy. "[I]t is worrisome that public energy-related R&D is declining in nearly every industrialized country", writes David Victor of the Council on Foreign Relations in New York, in an accompanying editorial in Nature. "In the United States – the biggest spender on energy R&D – the total funding fell 40% from 1985 to 1994".

This trend has been exacerbated by structural changes in the power-supply industry. Gas and electricity companies cut basic research by two-thirds from 1995 to 1996, "as restructuring and deregulation of energy markets led them to concentrate on short-term returns", writes Victor.

But Victor is sharply critical of a central feature of the latest research – the arbitrary choice of 550 ppmv of carbon dioxide as a suitable figure for stabilization. Why this number, and not another? "There is little solid evidence to justify that choice of number; nor is it clear whether any single number can express the damage that could result from global warming." Without some less arbitrary starting assumptions, says Victor, taking action on global warming is like "buying insurance with an unknown premium against unknown hazards."

Investing in new technology is a worthy aim, but if policy on global warming cannot offer clearer goals, it will be hard to persuade governments to spend so much, so soon, to mitigate disaster of unknown intensity, an unknown time in the future.