AI has been the dominant driver in markets this year. For Emerging Markets, this has played out in an extraordinary demand for those stocks that produce the ‘picks and shovels’ of the AI gold rush. Memory chip manufacturers have been in the vanguard of this rally. The recent correction has been a reminder that this historically cyclical business is still sensitive to the supply/demand dynamic. So where to next?
Clearly, if there is a reduction in the growth of AI capex committed, or if supply shortages of memory and all the other components of AI infrastructure are resolved sooner than expected, then a downturn in the cycle can be expected.
As to demand reflected by capex, the hyperscalers continue to ramp up spending with Jensen Huang, predicting that capex on AI infrastructure will be 4-5tn USD by 2030. The issue will be whether this type of investment is sustainable as financing shifts from existing cash-flow to debt funding. The Oracle downgrade is a red flag.
On the supply side the current consensus is that shortages will remain until at least 2H27, however, despite being denied the most advanced NVIDIA GPUs the Chinese pursuit of AI dominance could change the dynamic. The release of Moonshot’s Kimi 3 agentic model last week may be another DeepSeek moment and casts doubt on whether paying top dollar for the most advanced chips is money well spent. Chinese chipmaking companies are also a major factor when considering the supply side. The IPOs for CXMT (+460% on its first day of trading) and YTMC are a reminder of this, however, one analyst revealed that Chinese capacity would have to quadruple in order to satisfy domestic demand alone. The impact of Chinese production must also be viewed against the dominance of Hynix and Samsung when it comes to the most sophisticated chips.
The conundrum remains whether the correction just experienced marks a floor after all the excitement of the last 2 years or whether there is more to come. We have trimmed our AI exposure but still retain a decent position as the US hyperscalers are more likely to increase capex in the coming quarter than reduce it (as shown by Alphabet’s results this week). However, having a strong list of alternative options is important at this stage.
India, with its lack of exposure to the AI trade, has suffered from significant outflows from foreign institutional investors (FII) with 27bn USD of outflows as of middle of July and the lowest level of net FII buying since 2016. Clearly a fluctuating oil price may yet temper FII enthusiasm, however, it is reasonable to expect the rotation from India into the AI trade to reverse should AI start to show signs of fatigue. The first signs of this trend reversing have been visible with 1.8bn USD of net buying in the first fortnight of the month. Despite the lacklustre performance of the stock market of late, India is economically strong; growing GDP over 7% in each of the last two quarters and the stocks are significantly cheaper now than at the market’s peak in Sep’24. As you know, as stock pickers, we buy companies not markets and to this end we have added modestly to our Indian exposure albeit from low levels. Early days but on watch.
Click here to view the PDF for ‘The Impact Of The Tech Correction On EM Allocation’
