ROI not AGI (Guest: Leonid Mironov)

Leonid Mironov is a type nine licensed representative at Gavkal Capital in Hong Kong, working with Louis Vincent Gave on a new China fund that was not yet live as of May 2026. His core thesis is that sustained producer price deflation made Chinese corporate profitability structurally difficult, compounded by policy goals oriented around employment, technology, and unit output rather than profits, creating a self-reinforcing deflationary loop that reshaped buyer expectations across the economy.

China posted its first positive PPI print in February 2026, followed by a second at positive 0.5 percent in March 2026, which Mironov describes as acceleration rather than marginal improvement. His base case is that Chinese PPI inflation continues accelerating throughout 2026 as anti-involution policies filter through to consumer prices, though he characterizes the current state as green shoots in the numbers rather than in observable consumer behavior. Chinese CPI is distorted by specific moves including pork prices down 36 percent year to date for supply-side reasons. The anti-involution campaign comprises roughly 40 specific prescriptive measures rather than a broad declaration, reflecting a government style of consulting large manufacturers and small suppliers before legislating. The first concrete signal came around May 2025 when the government capped supplier payment terms for EV manufacturers at 90 days, ending practices like BYD's reported 360-day payment periods. After that shift EV makers stopped dropping prices, and the cheapest models introduced just before the policy at around 8,000 and 5,000 dollars represented a floor. At the Beijing Auto Show Mironov observed that newer EV models are better rather than cheaper. Q1 earnings across his monitored Chinese company dataset were up 16 percent, which he cites as evidence that profit focus is real against prevailing market skepticism.

China's net equity issuance turned negative for the first time on record in 2025, meaning companies paid more in dividends and buybacks than they raised through IPOs. Chinese households hold approximately 170 trillion RMB in savings with only 7 percent in equities, down from roughly 25 to 30 percent around the turn of the 2010s. Mironov argues rising inflation would redirect that capital toward equities and raises the scenario that China, widely expected to export deflation, could instead begin exporting inflation, warning few investors are positioned for that outcome. Stripping out AI and semiconductors, broad Chinese index valuations are approximately one standard deviation below historical averages, though those averages have been drifting down because valuations have been depressed for several years.

Mironov's central claim about Chinese AI is that no one in China is chasing AGI, with focus entirely on ROI from day one, driven by smaller lab budgets and a much thinner venture capital pool compared to the United States. Chinese AI applications tend to be narrow and specific, including manufacturing optimization, supply chain management, and autonomous mining trucks. Alibaba's Qwen and DeepSeek are notable foundational models but are framed as exceptions within a landscape dominated by application-specific products. Enthusiasm around embedded AI in Shenzhen collapsed when token usage bills arrived, with the same technicians who installed AI systems being rehired to uninstall them.

On oil, Mironov and the host assess the market as asymmetrically positioned to the upside, with roughly 15 dollars of downside if the Strait of Hormuz situation resolves and at least 30 dollars of upside, potentially reaching 150 dollars per barrel, if it does not. Polymarket odds as of late June put resolution at approximately 48 percent. The ProShares Ultra Short Oil ETF SCO grew from 126 million dollars in market cap before the conflict to 1.2 billion dollars, indicating the consensus short-oil-on-peace trade is already crowded. Even if trade through the Strait resumed immediately, the speakers argue it would take approximately six months to replenish global inventories and depleted strategic petroleum reserves, creating a structurally higher price floor. Iran's stated goal is characterized as keeping the Strait closed while avoiding being bombed, with both the US and Iran believing time is on their side, which the speakers identify as the core reason no resolution is occurring.

At the time of recording the S&P 500 was approaching 7,500 at all-time highs. BTIG data shows the NASDAQ 100 reached 13 percent above its 50-day moving average while touching a 52-week high, a combination that previously occurred only in September 2020 and at the March 2000 dot-com peak. The speaker's base case is a 20 to 30 percent semiconductor pullback while the S&P experiences only roughly a 5 percent correction from 7,500 to approximately 7,200, with money rotating into other sectors. Mironov is less optimistic, expressing concern that a semiconductor rollover could drag the broader market down rather than trigger a benign rotation. SK Hynix and Samsung together constitute approximately 50 percent of the South Korean KOSPI, with SK Hynix up roughly 400 percent in the year ending May, a concentration the speaker compares to Nortel and BCE comprising 30 percent of the Canadian index during the dot-com bubble.

Copper is breaking out of a flagging formation and reaching a fresh all-time high on a closing-price basis, with copper equities including Freeport-McMoRan, Ivanhoe Mines, and Southern Copper lagging the price rally. Mironov suggests copper stocks and gold miners would be preferred buys if the semiconductor bubble breaks.