“This year has been about deep value mean reversion, but next should see underlying secular growth themes reassert…” – Weekly Insight, 5th December 2016
‘Poker has been one of the hardest games for AI to crack, because you see only partial information about the game state. There is no single optimal move. Instead, an AI player has to randomize its actions to make opponents uncertain when it is bluffing…’ Andrew Ng, chief scientist at Baidu
Having been tactically overweight value (resources and industrial cyclicals) all of last year, we suggested in December that the post US election selloff in global tech as investors belatedly rushed to make reflation bets looked like an attractive contrarian opportunity. Since then both Chinese and US large cap tech names have outperformed from Alibaba to Facebook, with even Tesla enjoying a sharp rally. The overwhelming consensus expected a rampant USD, but this wasn’t consistent with the Trump trade agenda – the dollar index has had its worst January in over a decade. We remain tactically long EM local currency debt and short USDJPY but have now closed the long Japan exporter position for a 24% gain – Japan is still well represented in our long industrial automation and Embedded Intelligence sensor stock baskets. The dollar selloff has certainly helped the US tech sector given its high overseas exposure, but superior earnings momentum/visibility remains the key support, underpinned by broadly above expectations Q4 results.
At 22x earnings, the S&P 500 premium rating looks reasonable given a series of multi-year product cycles – the shift to cloud computing, rapid progress in AI/machine learning and a steady rollout of internet of things sensors as well as augmented/virtual reality offer sequential and overlapping growth drivers through end decade. The latest results from Microsoft, Intel and Alphabet indicated that cloud computing is the most significant near-term driver of growth while the recent CES hardware event in Las Vegas was dominated by new AI applications.
Meantime, a key theme remains that portfolio diversification will increasingly be embedded within 15-20 tech ‘conglomerates’ as they disrupt adjacent legacy sectors with lagging productivity. Amazon and Uber have both recently expanded into ocean freight and trucking logistics respectively. The key advantage these tech groups have is their ability to optimize slack capacity and automate scheduling via superior data algorithms. The latest developments in ‘reinforcement learning’ AI are remarkable, notably the Libratus program which has exceeded the much hyped achievement of Google’s AlphaGo by mastering the most complex form of poker. It learned the game from scratch using an algorithm called counterfactual regret minimization (which sounds like it might prove useful to many investors).
By playing at random initially, over several months of iterative ‘training’ (involving several trillion hands of poker) it reached a level where it could outwit elite human players by predicting their moves, playing a much wider range of bets and randomizing these bets – all without ever first being given the rules of the game. As we have long highlighted, the real long-term shareholder value in technology investing is in the layers of proprietary software that bind networks together and aggregate their data flow. The rapid progress in the models used for AI application development and associated processing hardware amplifies the scale of that opportunity (and also boosts the value of niche hardware sensor makers collecting the newly valuable data).
It seems inevitable that well paid ‘pattern recognition’ type roles from insurance claims assessors to junior M&A lawyers, auditors and hospital radiographers face automation over the next decade by increasingly intelligent software. AI technology has leapt with astonishing speed from facial recognition to putting on a poker face, and as it evolves further investors and politicians will struggle to absorb the profound economic implications, but one will be a growing concentration of incremental corporate earnings and free cash flow growth within the technology sector.