Policies designed to avoid economic downturns at all costs can lead to significant long-term risks. Capital and labor become trapped in inefficient companies that would otherwise fail, hindering productivity growth and creating a less dynamic economy.

When governments become top shareholders, corporate focus shifts from pleasing customers to securing political favor and appropriations. R&D budgets are reallocated to lobbying, and market competition devolves from building the best product to playing the policy game most effectively, strangling innovation.

JPMorgan CEO Jamie Dimon uses the CHIPS Act as a prime example of well-intentioned government policy becoming inefficient. He argues that while the core idea was good, it was diluted by a 'layer cake' of requirements from special interest groups, such as union mandates and childcare provisions, turning it into a 'swamp'.

OpenAI's offer to give the US government a 5% stake is not a public benefit but a strategic move toward cronyism. This would incentivize the government, as a shareholder, to create favorable regulations for OpenAI while over-regulating its competitors, effectively becoming a bailout disguised as an investment.

The U.S. is shifting from industry supporter to active owner by taking direct equity stakes in firms like Intel and U.S. Steel. This move blurs the lines between free markets and state control, risking a system where political connections, not performance, determine success.

Unlike the U.S. government's recent strategy of backing single "champions" like Intel, China's successful industrial policy in sectors like EVs involves funding numerous competing companies. This state-fostered domestic competition is a key driver of their rapid innovation and market dominance.

While investing in government-supported sectors like AI and semiconductors seems safe, it's a long-term risk. A government's priority is political—winning elections and preserving jobs—which will eventually conflict with an investor's goal of maximizing profit, leading to poor returns as seen in China.

Horowitz stresses that technological advancement is fragile. A single poor policy decision, like restricting GPU sales, can derail an entire industry and a nation's competitive advantage, regardless of its talent or culture. He points to a near-miss US executive order on GPUs as a stark example.

The US economy is now so dependent on the performance of a few AI-centric tech giants that their failure is not an option. When the AI bubble deflates, expect a government bailout, framed as a strategic investment like the CHIPS Act, to prop up the market and prevent a wider economic crisis.