Summary

The speaker’s base case is that the US will absorb the pain of tariffs and spending cuts first in 2025, while new tax cuts may not filter through until late 2025 or even 2026; investors should therefore cut exposure and hedge tail risk this year. The negative effects of tariffs and other policies are “happening right in front of us,” while the upside from new tax cuts will come later; in an extreme case, the US could enter recession before markets have fully priced it in.
If 25% tariffs on Canada and Mexico actually take effect, they will be far more destructive than the cumulative 20% tariff increase on China. Markets have already told Trump the policy is untenable through falling stocks, rising bonds, lower yields and a weaker dollar; the speaker still believes the Canada-Mexico tariffs may not last, with tariffs on China, critical imports and eventually Europe more likely to stick, though policy reversals themselves will amplify uncertainty.
Chinese assets are rebounding, but have nowhere near completed a bottom, let alone established an “East up, West down” regime. China’s PMI is above 50, property transactions are recovering, and southbound insurer flows and policy expectations can continue to lift Hong Kong stocks, but prices have not recovered; demographics, excess capacity and private-sector balance sheets remain three major overhangs. His sharp distinction is: “The process of a rebound is happier than bottoming, but the outcome of a rebound is uglier than bottoming.”
Trump could “mess things up” badly enough to push the US into recession in the short term, but the speaker remains bullish on America over the long term, with a recession potentially creating an entry point. DOGE’s cuts to government spending are like “scraping barnacles off a turtle”; after the short-term pain, returning resources to the private sector alongside sustained tax cuts could improve the long-term growth base. Meanwhile, as the US compresses demand, China—the largest supplier—will pay first.
AI “definitely has a bubble” in 2025 because investment is still rising while the efficiency gains needed to lift GDP will not be visible this year. DeepSeek has shaken the narrative that $1B-scale training clusters are indispensable, but the speaker believes cloud providers’ investment plans and NVIDIA earnings expectations have not fundamentally changed this year. The real test comes next year: whether investment can continue and whether breakthroughs on the capability side can enable tasks that were previously impossible.
The first area where AI can deliver tangible results is defense, followed by healthcare and transportation. The clash between Trump and Zelenskyy could push Europe to increase defense spending, while even if FSD and Robotaxi launches go smoothly in June, meaningful efficiency gains would likely not appear until next year. AI may not be a bubble over the long term, but “a GDP-scale payoff is definitely not coming this year.”
The speaker sees the promotion of a cryptocurrency reserve as a long-term negative, not credible institutional backing. If the goal were truly to build a national reserve, it should be announced by the White House rather than through an individual’s social-media account. He suspects it may share the same roots as the TRUMP coin operation, because “every penny it takes is money from the crypto world,” and says anyone participating should “believe early, leave early.”

Deep dive

1. The US and China are two sides of the demand-supply coin, but this is not “East up, West down”

The speaker framed the entire discussion as “A Tale of Two Cities”: the US once absorbed China’s supply, but as the relationship deteriorated, the US experienced inflation while China fell into deflation. Over the past 3 years, the US brought inflation under control and kept growth going through AI and technology, while China shifted toward a low-cost, manufacturing-heavy, intensely competitive “dividend society.”
DeepSeek changed the market’s view that Chinese technology stocks had “no seat at the table,” but he remains clearly cautious. His assessment is that roughly 20% reflects engineering improvements and 80% is “reverse-engineering the process from the result”; the comparison model may also have used OpenAI or another platform, so this is not proof that China has achieved a comparable breakthrough in original technology.
Hong Kong hedge funds have been debating whether China’s PMI breaking above 50 while the US PMI declines signals “East up, West down.” His answer: “East up, West down is absolutely not an option.” The more likely outcomes are “West up, East flat,” or “if the West does not rise, the East must fall.”

2. Once DOGE shakes government support, weak US data will translate into genuine recession risk

Over the past few years, US PMI and Consumer Confidence never became decisive recession signals. The key was government spending and credit growth: private-sector credit grew only 1%–2% of total credit, government credit grew about 10%, and the blended total still expanded by roughly 5%.
This year is different because DOGE could shake the government pillar. Core PCE has fallen from a plateau around 2.8% to 2.6%, while leading growth indicators are deteriorating. Major employment data such as nonfarm payrolls have not yet weakened, but initial jobless claims have exceeded expectations and do not yet fully reflect DOGE layoffs.
Markets have started discussing recession, but have nowhere near fully priced in that expectation. If aggressive tariffs and other policies actually take effect, the resulting adjustment could therefore be larger than current positioning implies.
The speaker does not view recession as the only possible outcome. If tariffs and deficit reduction ultimately prove to be “smoke and mirrors” and growth holds up, the Nasdaq could still rise 5%–10% for the year. But this would not be the “melt-up” of the past 2 years, so the return sacrificed by cutting exposure should be limited.

3. Tariffs hit growth immediately, tax cuts arrive later—the 2025 policy time lag is the biggest risk

The market’s standard bullish argument is that tariffs will weigh on the economy, but tax cuts can make up the difference. The speaker’s rebuttal is that the timing is asymmetric: the negative effects of tariffs and spending cuts will come first, while new tax cuts may not gradually filter through until late 2025 or even 2026. That is why “risk must be avoided in 2025.”
In his base-case estimate, if the effective tariff rate rises to roughly 4%, core PCE could increase by 0.4–0.5 percentage points and GDP could fall by about 0.5 percentage points—painful but manageable. But imposing 25% tariffs separately on Canada and Mexico, directly hitting the US’s 2 largest trading partners and potentially its third-largest trading partner, would be “extremely negative.”
Markets have delivered a clear immediate verdict on the policy: US stocks fell about 1.5%, bonds rallied, yields fell and the dollar weakened. The speaker therefore believes the Canada-Mexico tariffs probably will not remain in full effect. But China has already absorbed 2 rounds of 10% tariff increases in reality, so not every threat can be dismissed as negotiating smoke.
More likely to persist are tariffs on China, critical-import tariffs and potential tariffs on Europe. Even if Canada-Mexico or reciprocal tariffs are imposed first, they could be withdrawn after negotiations. His risk calendar includes a potential government shutdown on March 14, submission of a tax-reform plan to both chambers of Congress in April, legislation by the end of July, and roughly $100B in new tax cuts potentially not arriving until November. By comparison, spending cuts this year could reach about $150B.

4. China property and Hong Kong stocks have a tradable rebound, but deflationary constraints remain

China’s property transactions in January and February posted their most meaningful recovery in roughly a year, with second-hand homes accounting for about 40% and volumes also recovering. But the speaker emphasized that “prices have not recovered; transactions have,” which is not a synchronized repair of prices and balance sheets.
The rebound also has clear funding and policy fuel: money printing and increased government-bond issuance began last September, policy expectations around the National People’s Congress strengthened, and ICT hardware investment began shifting toward government procurement of software and AI software. Southbound flows include substantial insurer allocations; he has even heard that PBOC funds are allocating as well.
The improvement in trade may reflect inventory restocking. In the US, companies built inventories ahead of tariff implementation, while the speaker estimates that roughly half of the drivers attributed to China’s GDP growth last year came from foreign trade, with final demand still mainly coming from the US. After the cumulative 20% tariffs on China flow through over several months, trade data could easily weaken.
Debt resolution can push liabilities into the future through government money printing or additional leverage, but that does not repair household and corporate balance sheets. Demographics, excess capacity, cash-strapped households and a private sector unable to add leverage all ultimately point to the same problem: “There is no consumption.” Hong Kong financing and A+H listings are easier to approve, so they may serve financing and debt resolution rather than signal a reversal in the earnings cycle.

5. Chinese technology valuations can expand, but earnings remain capped by macro demand

DeepSeek lowered the floor for many software companies, while government procurement and sentiment could support a months-long rally in Chinese TMT. But the speaker cautioned that as Chinese TMT companies’ market caps have risen, the earnings of each customer—the “UP value” in his terminology—are still falling. Weak macro demand makes it difficult for profitability to recover.
Even if the US implements both supply-side and demand-side reforms, manufacturing would struggle to return immediately: when demand falls, policy cannot quickly bring domestic supply online, and the largest existing supplier gets hit first. He summarized the chain reaction as: “Trump messes things up, and China pays the bill.”
After being bearish on China for 3 years, he had expected a roughly 6-month rebound. It did not happen last year, but policy changes could make it happen this year. The characterization must remain precise: “This is not bottoming; it is a rebound.” After every rally, the downtrend still has little resistance.

6. AI must prove its valuation through productivity, while crypto has exposed the fragility of policy narratives

In his view, AI’s impact on the broader industry and the stock market is currently showing up mainly in NVIDIA. Massive investment has driven Nvidia’s share price, while OpenAI is the “front-end advertiser.” DeepSeek destroyed the certainty that everyone must build enormous training clusters, leaving the market unsure how to value NVIDIA or where the eventual capex should go.
Online reports of CoreWeave order cuts do not necessarily signal a deterioration in real demand. At the beginning of the year, companies often have to overstate demand to secure CoreWeave’s available capacity; subsequent reductions do not mean that actual demand or the data used in companies’ earnings forecasts has changed. He believes this year’s CoreWeave orders and overall capex investment have not been affected. The biggest test remains whether cloud providers can continue adding investment next year.
A genuine breakthrough on the capability side is not more investment; it is efficiency gains that lift GDP and enable tasks that were previously impossible. Defense can be the first application. The clash between Trump and Zelenskyy could also push Europe to increase defense spending. Healthcare, FSD and Robotaxi will take longer—even if progress in June goes smoothly, meaningful results are unlikely before next year.
On crypto, he refuses to treat the reserve announcement from Trump’s personal account as White House policy and even suspects that his team and the team operating TRUMP coin may be the same group. Every $100M pulled out leaves $100M less in the pool. Friday’s meeting may produce a short-term bounce, but his own positioning remains somewhat bearish.

Devon Zuegel: How To Create A New Town - [Invest Like the Best, EP.413]How Claude Plays Pokémon was made