Just finished watching the latest episode of the All-In Podcast. The midterm elections saw a sweeping Democratic victory, $4 trillion in AI stocks evaporate in a single day, and OpenAI overturned a fundamental assumption in computer science. Midterm Election Recap: Democrats Sweep Both Houses, Traditionally Red States Unexpectedly Turn Blue In this midterm election, the Democrats won majority in both the House and Senate, and also won several gubernatorial seats. Traditionally red states like Texas and Kansas unexpectedly turned blue. The four hosts analyzed the root causes of the Republican defeat from their respective perspectives. Sacks believes the root cause of the defeat lies in the Iran war and its ensuing economic consequences. The war caused oil and diesel prices to skyrocket, pushing up prices and food costs. The rebound in inflation completely dashed the Federal Reserve's expectations of interest rate cuts, and mortgage rates returned to high levels, directly destroying affordability for ordinary people. He also mentioned that the war broke the political alliance between the Republican Party and young male voters and the independent podcasting community (such as Joe Rogan and Tucker Carlson), resulting in a significant loss of these voters. Friedberg's focus is on the deeper issue of fiscal imbalance. He emphasized that the core issue lies in excessive government spending. After the Department of Government Efficiency (DOGE) reforms stall and are terminated by mid-2025, trillions of dollars in annual deficit spending directly drive up costs across society. Ordinary citizens feel no benefit from the prosperity, and this disparity prompts voters to cast their ballots for candidates advocating government intervention and redistribution. Chamath pointed out that ideological labels are no longer important in this election; the core issue is the breakdown of the underlying mechanisms of the American economy. In the past few years, regardless of changes in the macroeconomic environment, the wealth of asset holders has continued to increase, while the actual purchasing power of the working class has been severely deprived. This voter anger is essentially a retaliatory vote against economic inequality. Jason believes that voters abandoned the ruling party because Trump broke his campaign promises to address inflation and not start new wars. Coupled with inflationary fluctuations caused by tariff policies and law enforcement controversies, his support among centrist voters has plummeted. Several Self-Help Suggestions for Trump's Second Half of His Term Regarding how to proceed in the next two years, several hosts also offered some quite specific political self-help plans. First, end the war as quickly as possible to lower international oil prices, leading to a decline in fuel, transportation, and food prices. This would create room for the Federal Reserve to cut interest rates, fundamentally alleviating the affordability crisis for the public. Second, shift to a pragmatic, transactional approach to governance. Abandon pointless partisan infighting and, while preserving the achievements in border control, make trade-offs with congressional Democrats on some populist economic policies, such as raising the minimum wage, exempting low-income individuals from federal taxes, stimulating new housing construction, and restricting public officials from stock trading. Sacks made an interesting historical analogy. Clinton in 1994 and Obama in 2010 both won re-election after devastating midterm election defeats through policy adjustments. He cautioned against repeating the mistakes of George W. Bush in 2006, when his all-in troop buildup in the Iraq War ultimately led to a larger-scale political collapse. $4 trillion in AI stocks evaporated in a single day; data center approvals are being halted by local politics. Market concerns about a halt in computing infrastructure construction caused nearly $4 trillion in market capitalization to evaporate in a single day, also dampening expectations for the IPOs of Anthropic and OpenAI. The core conflict centers on local political resistance to large data centers. Local communities and the newly elected governor are strongly protesting, arguing that hyperscale data centers are driving up electricity bills and consuming vast amounts of water resources, while they themselves enjoy huge tax breaks. This antagonism—that tech giants profit while local residents foot the bill—is escalating. Chamath conducted a crucial industry landscape analysis. Slower approvals will severely impact emerging cloud service providers and cutting-edge infrastructure companies that rely on leased computing power. Meanwhile, hyperscale vendors with their own energy and infrastructure reserves (Amazon, Google, Microsoft, Meta, SpaceX) will control the pricing and allocation of scarce computing power. This political turmoil is likely to accelerate industry fragmentation, with the strong becoming stronger. Jason and Sacks also warned of a longer-term risk. If states across the US implement strict regulations or freeze approvals, data centers and computing infrastructure will be forced to relocate to Canada, Mexico, Southeast Asia, or the Middle East. This will take away not only high-paying engineering jobs but also corresponding tax revenue. The US could potentially transform from a net exporter of computing power into a net importer of tokens. While this statement is somewhat exaggerated, the direction is one to be wary of. Science Corner: OpenAI Overturns a Fundamental Assumption in Computer Science At the end of the program, Friedberg introduced a theoretical breakthrough. The exponential time assumption (ETH), since its inception in 1999, has been a key cornerstone in theoretical computer science for determining that NP-complete problems (such as 3-SAT) cannot be solved in sub-exponential time, and a core pillar in exploring the age-old problem of whether P equals NP. OpenAI recently published a 175-page paper proving, through a completely new methodology, that the 3-SAT problem can be solved in sub-exponential time, mathematically falsifying ETH. This means that all problems in the universe that can be quickly verified may be efficiently computed (i.e., the possibility that P=NP is reactivated). This breakthrough not only demonstrates the disruptive potential of AI systems in solving highly difficult pure mathematics and fundamental scientific problems, but also triggers a comprehensive re-examination of future theories of computational complexity within academia.