Regular gasoline cars originally relied mainly on naturally aspirated engines. Later, small-displacement turbocharging technology expanded massively, almost driving natural aspiration out of the mainstream gasoline car market. However, as hybrid vehicles became popularized, many new hybrids, including those from Toyota, returned to naturally aspirated engines.
This boomerang-like technological choice indicates that the automotive industry does not always evolve toward unidirectional technical metrics. What determines a winning route is often not which engine is forever more advanced, but how the external environment alters the optimal solution for the entire system.
The critical variable determining this optimal solution is the cost structure. When fleet carbon dioxide emission regulations attached a clear price to emissions, the already viable turbocharging technology crossed the tipping point for the mass market. Subsequently, concentrated orders brought about scale effects, manufacturing learning, and the gathering of supporting suppliers, further reinforcing this choice. Mobile phone CMOS, electric vehicles, the return to natural aspiration in hybrids, and even today’s artificial intelligence competition all demonstrate different facets of this mechanism within their respective fields.
How much power an engine can output per combustion depends first on how much air can fit inside the cylinder. A naturally aspirated engine lacks a dedicated compressor and relies solely on the downward movement of the piston to reduce in-cylinder pressure, allowing the external atmosphere to push fresh air into the cylinder. The only initial pressure engineers can use is the surrounding atmospheric pressure. To pack in a bit more air, they calculate the length and cross-sectional area of the intake manifold, coordinating with the timing of valve openings, in an attempt to use the reflected waves of airflow to press more oxygen inside.
Turbocharging changed the intake method. It uses the expelled high-temperature exhaust gas to drive a turbine, which spins a coaxial compressor, directly increasing the density of the air fed into the cylinder. Small-displacement engines thus gained higher power density and mid-to-low RPM torque. But this technology introduced new engineering compromises. It takes time for the exhaust gas to spool the turbine, causing a lag in power response. The system also has to bear complex plumbing designs, higher knock risks, and severe cooling and material tests.
If other constraints were not considered, natural aspiration and turbocharging could have easily coexisted long-term in their respective areas of expertise. A retrospective on Golf history by Volkswagen notes that the 2004 Golf GTI featured a turbocharged gasoline engine, and a TSI twincharger version was introduced in 2006. The European new car data compiled by the ICCT also shows that turbocharged products were already in mass production, and the market share of European gasoline turbo models had begun to rise before 2009. But merely looking at the gains and losses in technical parameters cannot explain why turbos later became the nearly default configuration for regular gasoline cars in the mass market.
Europe’s carbon emission regulations intertwined with the expansion timeline of small-displacement turbos. Automakers had voluntarily committed to reducing carbon dioxide emissions in 1995, but because the targets were clearly unmet, the European Commission announced plans for mandatory regulation in 2007 and officially finalized the regulations in 2009. The ICCT’s review of the European policy process documents this journey.
The newly enacted Regulation (EC) No 443/2009 did not mandate which technologies automakers should adopt, but instead evaluated the average carbon dioxide emissions of new cars sold by each automaker that year. Once the fleet average exceeded the limit, automakers had to pay fines based on the excess grams multiplied by the total number of vehicle registrations for the year. Between 2012 and 2018, the first three grams over the limit corresponded to fines of 5 euros, 15 euros, and 25 euros per vehicle respectively, rising to 95 euros from the fourth gram onward. Starting in 2019, every excess gram was calculated directly at 95 euros per vehicle.
Multiplying fines by sales volume changed how engine technologies were evaluated. Small-displacement designs could reduce a portion of pumping and friction losses under low loads, thereby effectively lowering emission figures in official test cycles. When this fuel-saving effect was converted into the financial value of avoiding massive fines and entered automakers’ ledgers, the relative economics of the relevant technologies shifted, making turbocharging much more attractive. As discussed in Disposable Software and Compressed Reality: The Essence of AI Native is Strategy Reconfiguration, best engineering practices are often products of a specific price table. Regulations placed a clear price on emissions, and a solution that originally lacked absolute advantage at the system level became a rational strategy to help automakers avoid exorbitant expenses.
After a technology crosses the mass market threshold driven by an external push, automakers and suppliers typically do not re-evaluate all technical routes every year. The production and investment accumulated in the past will continuously lower the cost of the current solution.
This cost advantage is maintained by several distinct mechanisms. First, as production increases, the fixed costs of upfront product research, factory construction, and custom tooling distributed across each engine drop significantly, which is the economy of scale. Second, as cumulative production grows, factory teams master the process, manufacturing yields improve, and raw material waste in workflows decreases, a phenomenon known as learning-by-doing. Meanwhile, technological advancements independent of production scale, such as combustion chamber architecture evolution and control algorithm optimization, also provide gains. Furthermore, when small-displacement turbos become mainstream, supporting capital, testing equipment, and engineering talent naturally cluster around them. Once these assets are committed, if other routes wish to compete again, the experience and support networks accumulated by the incumbent route become a steep replacement cost.
As the image above illustrates, establishing an advantage relies on sequential transmission across stages. After external regulations or large market demands concentrate, a technical solution needs to precisely fit these constraints to capture orders. Secured orders bring scale effects that dilute costs and build learning experience in manufacturing, which in turn attracts supporting investments. However, this cycle also has boundaries where it can fail. Concentrated demand does not mean the supply chain will successfully handle it, and capacity expansion does not guarantee an automatic improvement in factory yields. If the product loses touch with users’ genuine usage needs, or if industry capacity expands excessively, companies will still struggle to turn a profit. Only when these stages interlock can an initial market share advantage translate into a long-term competitive moat. Outside of policy, the mobile phone market’s selection of CMOS sensors provided a different kind of initial pull.
Around 1990, CCD was already the technically mature, high-image-quality industry mainstream. Early CMOS image sensors lagged in noise control and pixel consistency. Competing purely on image quality metrics, it lacked an advantage.
However, CMOS APS technology had a critical trait suited for emerging electronic devices. It could integrate circuits for signal amplification, analog-to-digital conversion, and image processing directly onto the same silicon die. This high level of integration drastically reduced the footprint and power consumption of the camera system, and it lowered the difficulty of peripheral circuit design.
For the newly emerging camera phones, the primary constraints were precisely small size, power efficiency, and cost control. Camera phones became the killer application for CMOS image sensors. Massive orders attracted capital inflows, propelling a rapid expansion in CMOS production scale. By 2013, the global annual production of CMOS sensors surpassed 1 billion units. Scale diluted manufacturing costs and provided manufacturers with the budget to resolve early flaws like noise.
In the latter half of this process, Sony played a late-stage accelerating role. The column-parallel analog-to-digital conversion technology it developed improved reading speeds and suppressed noise, and the subsequently introduced backside-illuminated structure and stacked design further enhanced light sensitivity. While Sony did not invent this technology, this evolution demonstrates the typical pattern of crossing a tipping point. Massive demand first filters for a technology that fits the core constraints, and subsequent capital investment and manufacturing learning help patch up its initial shortcomings.
In their early development, electric vehicles needed to establish new powertrain and battery supply chains. With hefty upfront investments paired with limited early demand, it was difficult to complete a cold start relying purely on the commercial market. The International Energy Agency’s summary of early EV policies records how policy broke the supply-and-demand deadlock. Consumer-facing subsidies and tax incentives directly reduced purchase prices, while fleet average emission standards and Zero Emission Vehicle (ZEV) rules aimed at automakers increased the compliance value of supplying EVs.
After policies helped establish the initial market, production experience began to play its role. A revised NBER working paper used structural modeling to estimate that, within its research sample, after controlling for variables like economy of scale, overall industry technological progress, vehicle assembly experience, and raw material prices, every doubling of cumulative experience in battery production lowered unit production costs by 7.5%. While this is only an estimated figure from a specific model, it quantifies the persistent impact of manufacturing experience on subsequent production costs.
As the market developed, sales data shifted noticeably. In China’s domestic passenger car sales in 2025, the share of New Energy Vehicles (NEV) reached 54%. New Energy Vehicles here include Battery Electric Vehicles (BEV), Plug-in Hybrid Electric Vehicles (PHEV), and fuel cell vehicles. In the United States, Washington state’s 2030 transition target does not carry a direct mandatory order; what is truly legally binding is the requirement that by 2035 all new passenger vehicles sold must meet zero-emission technology standards. Under this mandate, new plug-in hybrid electric vehicles that meet the standards can still be sold, and the use and trade of used gasoline cars remain unrestricted.
However, simply expanding capacity does not inevitably bring sustainable cost advantages. According to 2024 statistics from the International Energy Agency, global nominal annual manufacturing capacity for electric vehicle batteries reached 2.2 TWh in 2023, while actual battery demand was only 750 GWh. If paper capacity lacks the backing of real market demand, idle equipment can neither guarantee improvements in production yield nor ensure sustained profitability.
The image above compares the paths taken by turbos, EVs, and CMOS. CMOS fit camera phones’ requirements for power consumption and integration, completing a spontaneous launch in the commercial market. Turbos and EVs relied more on the push of regulatory oversight, crossing early thresholds by altering manufacturers’ compliance costs. After clearing the tipping point, all three entered a reinforcement phase that relied on order scale to dilute costs and on manufacturing learning and supporting investments to consolidate advantages. At the same time, if a product detaches from real demand, or if production capacity expands too much, this scale-dependent cycle also faces the risk of failure.
Policy support not only propelled battery electric vehicles but also helped hybrid technology expand its market share. As electric motors became part of the powertrain, the division of labor for propulsion inside the vehicle shifted.
Electric motors possess robust low-speed torque and swift power response, capable of sharing the load during vehicle launch and transient acceleration. The engine therefore no longer has to frequently deal with inefficient operating conditions and can operate stably in highly efficient RPM and load zones more often.
This new division of labor altered the system yields of different technologies. The power density and mid-to-low RPM torque brought by turbocharging were once its core advantages, but the addition of electric motors shared the demand for low-RPM torque, making this contribution from turbos no longer as scarce. Meanwhile, the complex plumbing, thermal management burdens, and power lag of a turbo system still remain. By contrast, a naturally aspirated engine paired with a specific high-efficiency cycle appears far more reasonable in the cost and benefit trade-off of the entire system.
Mass-produced hybrid engines often adopt the Atkinson cycle. By delaying the closure of intake valves, this design makes the expansion ratio greater than the effective compression ratio, thereby reducing the heat carried away by exhaust gases. This results in the loss of some mid-to-low speed torque, but in a hybrid system, this exact shortfall is compensated for by the electric motor. Technical data published by Toyota in 2014 showed that its 1.3-liter naturally aspirated engine achieved a maximum thermal efficiency of 38%. A subsequent paper on the Dynamic Force powertrain noted that the conventional version of its 2.5-liter naturally aspirated engine reached a maximum thermal efficiency of 40%, while the dedicated version used in hybrid systems peaked at an impressive 41%. These figures refer to the peak thermal efficiency of the engine block itself, not the overall vehicle efficiency, but they illustrate that naturally aspirated technology has found a new role in tandem with electric motors.
The return of natural aspiration in the hybrid era demonstrates a secondary reorganization of the cost structure. The introduction of new propulsion components changed system constraints, and an old technology that once seemed marginalized can still return to a core position through system-level cost advantages, provided it adapts to the new role division.
This law of cost structures determining technical solutions applies equally to today’s artificial intelligence competition. When evaluating AI models, the tech circle often focuses on capability evaluation metrics (benchmarks). However, determining which architecture can secure enterprise procurement budgets and become the mainstream market solution is similarly subject to changes in the cost structure.
When compliance requirements, industry standards, and security liabilities draw hard boundaries, the basis on which enterprises evaluate technical routes also shifts. Recalling the previously mentioned Disposable Software and Compressed Reality: The Essence of AI Native is Strategy Reconfiguration, the optimal strategy is a product of a specific price table. If regulatory policies or procurement rules demand that a system must provide output auditing, guarantee local data residency, or include human sign-off steps, these engineering options that originally seemed peripheral will become prerequisites that enterprises must satisfy to purchase services.
In such an environment, a technical solution that can prioritize solving audit compliance and workflow integration costs will win orders faster than a competitor solely focused on singular model parameters. But this is merely the first step across the tipping point.
Whether initial orders can translate into lasting advantages depends on whether companies can navigate the subsequent production learning curve during delivery. If, in the process of delivering to more clients, the system’s deployment cycle does not shorten, unit compliance costs do not drop, error rates do not improve, or clients jump ship once policy or procurement preferences wane, then the market share initially gained through compliance requirements will be short-lived. Only by rapidly converting early orders into more efficient deployment tools and standardized delivery workflows can a technical route maintain its competitiveness through accumulated engineering efficiency, even after policies or procurement preferences weaken.