Turbines Make the World Go Round

February 2026

Turbines have built a modern civilisation dependent on controlled rotational energy, and have become the cornerstone of our industrial world. They have gained prominence amid the AI hype as gas-powered generators have become the main workhorses for rapidly scaling power generation for data centres, and have subsequently become a key bottleneck. The role of turbines, reductively, is to extract useful work from ordered energy flows (typically the expansion of high-pressure fluids) and convert thermodynamic potential into controlled rotation. At the grid-level, they convert large scale energy flow into a controllable rotational motion geared for electrical generators, aiming to do as efficiently as possible. That could be the movement of water under gravity, or of steam and gas under pressure.  I will be focusing on common and mainstream turbine systems involved in electrical generation for the grid. That means focusing on heat engines converting from fuel (steam, gas, combined-cycle) and flow harvesters capturing already mechanical energy (hydropower and wind), and not geothermal, marine/tidal, supercritical CO2, or ORC. 

  • Steam turbines work on the Rankine cycle, which expands high pressure steam through many stages to maximise enthalpy extraction while maintaining blade efficiency. It underpins coal, nuclear, and biomass. ~40% efficiency. 

  • Gas turbines work on the Brayton cycle. They compress air, burn fuel, and expand the resulting hot gas through a turbine that drives the compressor as well as a shaft for electrical generation. Unlike Rankine cycles, Brayton cycles operate under continuous combustion and flow. ~40% efficiency, limited by the demands of the compressor, the limits on inlet temperature (~1500°C), and the high outlet temperature (~600°C). 

  • Combined-cycle gas turbines (CCGT) takes advantage of both the Rankine and Brayton cycles. A CCGT starts by generating power through the Brayton cycle, from which the ~600°C exhaust heat is recovered to generate steam which in turn drives the Rankine cycle. This drives a higher ~60% efficiency. 

  • Hydropower turbines are driven by the movement of water under natural flow or gravity. ~90% efficiency, helped by water’s 800x density over air but lessened by flow separation and turbulence, friction and cavitation (where water locally boils into vapour). 

  • Wind turbines likewise are driven by the flow of air. ~50% efficiency, limited by the Betz limit which identifies that for power to be extracted from wind, downstream windflow cannot be zero to sustain continuous mass flow. 

Note that, generally, all of the above use axial turbines (fluid moving parallel to the axis of rotation) rather than radial turbines (fluid moving perpendicularly to the axis of rotation) because for grid-level production axial turbines scale efficiently with mass-flow, allow the stacking of multiple blade stages, and rotate with lower relative centrifugal stresses. You’ll also note that the first three turbines are limited in efficiency as heat engines by the Carnot limit, which states that the maximum theoretical efficiency of converting heat into work is limited to the extent of the temperature gradient. In reality, they don’t get close to that limit due to material limits and compressor work. 

The 21st century will see a shift in the role of turbines from one of thermal extraction (Brayton/Rankine/CCGT from coal, nuclear, biomass) to flow extraction (wind and hydro) thanks to the latter’s lower marginal cost and higher efficiency. Traditional use cases where steam and gas turbines dominated will give way to the rise of capturing natural flow states in air (wind) and water (hydro). Turbines will also become bigger. Combined cycle gas turbines are now at 500MW at the heavy-duty scale, and are at ~65% efficiency. Wind turbines have noticeably grown with some reaching 250m rotor diameters. This is because power scales with the square of diameter, while the balance-of-plant costs, maintenance, grid infrastructure and construction, scale approximately linearly. 

 

There are two trends at play here to understand the scale of turbines’ contribution to world energy and economic systems. One is the decline of turbines as a share of global electricity production, with fossil fuel power stations showing the most decline and solar PV growing the fastest (more on that later). The other is the growing electrification of world energy supply. 

The world’s annual energy generation of 180PWh can be broken down to: 
Oil (30%), coal (25%), natural gas (25%), biomass and renewables (15%), nuclear (5%)

The world’s annual electrical generation of 30PWh can be broken down to: 
Coal (35%), natural gas (23%), hydropower (14%), nuclear (9%), wind (7%), solar (7%), other (6%)

90%, 27 PWh,  of global electricity is generated by the rotation of turbines, with the one major exception being solar PV.  In total though, less than a fifth of global energy is electrical by source (30PWh / 180PWh = 17%). This proportion is of course rapidly accelerating as the world electrifies. Electrical systems are generally far more efficient, at around 90% roundtrip efficiency vs anything from 20%-80% for fossil fuel-powered systems (lower for motion, higher for heating). Electricity production is subject to more favourable learning curves thanks to lower marginal cost renewables and is less carbon-intensive.

By 2100 electricity generation is predicted to 8x to around 240PWh, and global energy production to 2.2x to around 400PWh. 240PWh / 400PWh = 60% of global energy supply will be electrified.

Of this 240PWh of electrical generation, 35% will be solar, which leaves 160PWh of turbine-generated power by 2100, a 6x increase from now. So even as solar drives a reduction in turbine-generated electricity by proportion, total electricity demand grows so much that there is still a dramatic increase for turbines in absolute terms. 

 

Solar is the white elephant in the room here. Photovoltaics has been the darling energy story for techno-optimists in the last decade. It is the fastest growing energy technology ever, with installed capacity going from 1GW to 1TW in just two decades. Besides revolutionising grids on earth for the better, with cheap, low-maintenance, and carbon-free electricity, it may also underpin a whole array of futuristic opportunities from space-based data centres and factories, to modular synthetic fuel labs. But at grid-level, solar’s cheap marginal cost of power production clashes with its increasing cost to scale for grid integration. As solar scales to a plurality of a grid’s production (about 30-50%), it places ever an ever-growing burden on battery requirements, transmission infrastructure, and grid inertia. BESS can somewhat counter the problem of grid inertia, but they are not instantaneous and require a mesh of active control, software and system dependence. Wiring up BESS to the grid to counter solar partially solves inertia but complicates grid transmission infrastructure non-linearly. 

This is not an anti-solar diatribe. Far from it. This is to recognise the complementary offerings of the two energy technologies; namely that solar has economies of scale in the marginal cost of generation, but has diseconomies of scale in grid integration. Turbines do not (yet?) have economies of scale in the marginal cost of generation but have economies of scale in grid integration. 

Solar’s limits are dominated by physical factors, namely inertia and inherent supply/demand mismatches driven by intermittency. Turbines limits are material science bottlenecks, which may be a physical limit or merely a manufacturing limit. If the latter can be overcome then humanity will become much wealthier, especially those responsible for commercialising it.

 

Our earlier estimate projects a ~6x increase in turbine-generated electricity from 27PWh to 160PWh by 2100. This understates turbines’ true impact on the economy. As a heuristic, electricity has approximately 3x the contribution to economic activity vs non-electrical energy due to roundtrip efficiency and enabling mechanical and informational precision. AI may radically increase this as it consumes the digital and perhaps physical world, but we retain a 3x multiplier to be conservative, recognising that the marginal economic contribution of energy also tends to decrease as economies mature. 

At 17% of energy generation today, we can hypothesise that electricity drives about 38% of global economic output with this 3x multiplier. With the world’s GDP nominally $110tn, that is $42tn of economic activity supported by electricity. Turbine-generated power is upstream of 90% of this as mentioned, which is $38tn of economic activity. 

Let’s roughly estimate world GDP to be $500tn by 2100. I’m aware of the scope of variation in that, but directionally it is useful. At that point, turbines will be producing ~65% of electricity, which itself will power 60% of global energy use at the same 3x economic premium, giving $260tn worth of economic activity that is downstream of turbine-generated power. 

So between now and the end of the 21st century, turbines will 6x in electrical output, and 7x in supported economic activity.

 

How turbines are made is a whole separate piece. The bottom line is that they are very hard to make reliably and economically at scale. Brutally hard, in fact. There are only three tier-1 companies on earth that can do so for gas and steam turbine thermal power plants: GE Vernova, Siemens Energy, and Mitsubishi Heavy Industries. Hydro and wind are slightly less concentrated, but still operate as businesses with extremely high barriers to entry. In essence, the limiting factor is the frontier of material sciences. Decades of cutting-edge metallurgy have incrementally improved the tolerances of alloys that operate at temperatures above the melting points of metal. Turbines must operate for hundreds of thousands of hours, complete tens of billions of rotations, and do so with minimal risk of breakdown or downtime lest they cause immediate danger or shut down electrical supplies costing millions of dollars per minute. They must not only operate without failure but do so to the absolute limits in order to drive ever higher efficiencies. Turbines operate over multi-decade iteration cycles, and cost billions of dollars to develop.  

A turbine whirrs faraway in the distance to power the device you read this on, while the world’s economy rotates evermore around its shaft.