Commodity prices turn on a nickel — a metal, incidentally, worth three-fifths what it was a year ago. This is a sector both cyclical and in hock to all sorts of prevailing winds: geopolitical, economic, technological and Mother Nature herself. Is it time to turn price discovery over to artificial intelligence?
The US thinks so, at least when it comes to critical minerals such as nickel, lithium and a swath of the periodic table. The Department of Defense wants to use AI to plot the supply and demand of critical minerals and shine a light on what it sees as the current opaque pricing.
To this end, it has charged the Defense Advanced Research Projects Agency (Darpa) with spurring tech advances “to enable a more reliable signal for buyers, sellers and investors than is currently available”. Access for retail investors is unconfirmed, with the project recently put out to tender. But it could be a useful tool for those with commodities exchange traded funds.
The Pentagon’s interest in the opaque world of materials pricing is understandable. There are both industries and nations to protect. Lanthanum is used in night-vision goggles; molybdenum in missiles; silver in Apache helicopters; nickel in body armour. Much of this is imported: more than half of all but two of 12 key critical minerals for the US aerospace and defence industry, reckons the Aerospace Industries Association.
Nor is the US the only country fretting. Deglobalisation was already well under way before Russia’s invasion of Ukraine illustrated the costs of being beholden to a country for essentials like fuel, grain and fertilisers.
Harnessing AI to predict supply and demand and estimate commodities pricing tops off a timeline that began maybe six millennia ago when folk traded livestock and shells in ancient Mesopotamia.
More recent iterations, namely the Chicago Mercantile Exchange and London Metal Exchange, have had their share of snafus. Recall a couple of years back, when the LME’s response to a spike in nickel prices — squeezing short positions — was to halt trading and cancel transactions, leaving the market in disarray.
The ability to capture such exposures would guard against chaos. But this is not a particularly big ask of exchanges themselves. Stock exchanges, for example, manage to do so.
AI captures the hypegeist, but like any existing models relies on data inputs. Big data in the commodities world is partially cloaked and comes with manifold variables like transport (including canal blockages and pirates on the high seas) and storage costs.
There is a limited, if admittedly growing, data set on zigzags and U-turns in policy and technology. There is the waning appetite of traditional carmakers like Ford and rental outfits such as Hertz for battery-powered vehicles. Even assuming EVs regain their momentum it is closer to crystal ball work to nail down what elements will power them. Layer on to that governments’ backtracking on their commitments to the energy transition.
One straightforward input — whether on AI, spreadsheets or an abacus — should be stockpiles. True, these will become more transparent as countries, including the UK, and the International Energy Agency address vulnerabilities in supply.
This month the IEA unveiled plans to secure enough minerals such as lithium, cobalt and copper, much as it does with oil. Under that programme member countries are required to hold 90 days of net oil imports which they can release in the event of severe supply disruptions.
But there is a huge wild card in China, which dominates in production as well as refining and processing of many of these. It has been building reserves of the likes of cobalt, as well as acquiring mines across the globe. But forget about transparency. Resolving that particular informational asymmetry challenge is likely to prove one of the stiffest tests for Darpa’s planned AI model.
Bitcoin ETF boost cuts two ways for Coinbase
Coinbase investors should be in celebratory mood. The crypto trading platform last week posted its first quarterly profit in two years. Trading volume bounced back towards the end of last year as a bitcoin rally and expectations that regulators would approve the creation of the first spot bitcoin exchange traded funds triggered renewed interest in the token.
Having surged nearly 400 per cent in 2023, shares in Coinbase are about flat this year.
The excitement was premature. While the Securities and Exchange Commission gave the green light to 11 bitcoin ETFs in January, the long-term benefits to Coinbase are not clear cut. Spot bitcoin ETFs offer US investors exposure to the world’s largest cryptocurrency without directly holding it.
The funds, which include those from BlackRock, Franklin Templeton and Invesco, have attracted net inflows of nearly $4bn, according to ETF.com.
Coinbase stands to benefit by acting as the custodian for the bitcoins held by these funds. Although the $69.5mn it made from custodian fees in 2023 accounted for just 2.2 per cent of total group revenue, the figure should rise this year. The company has said that it serves as custodian for eight of the 11 spot bitcoin ETFs.
The downside is that margins from custody are thin. Investment bank Mizuho reckons Coinbase earns a fee of around 0.07 per cent for its service. That compares with trading commissions of as much as 0.6 per cent that it can charge customers every time they buy or sell cryptocurrencies.
The bigger worry is that the rise of low-cost spot bitcoin ETFs will give investors fewer reasons to trade actual bitcoins. For Coinbase, a volume drop in the trading business could quickly offset any gains from the custody of assets.
So far, Coinbase says it has not seen evidence of cannibalisation. First-quarter subscription and services revenue could grow by as much as one-third year on year to $410mn-$480mn, the company said.
It is not clear how much of that growth is coming from interest income that it earned on its stablecoin reserves and other products. This has become an important revenue stream, growing 150 per cent last year to about $868mn — or 30 per cent of total group revenue.
Coinbase’s valuation looks full. The shares trade on 14 times revenue, up from three times a year ago. That is more than twice the multiple for Robinhood. Larger and more profitable exchange operators Cboe and the London Stock Exchange are on five and six times respectively. Another reason, then, for investors to avoid getting swept up in crypto’s ETF frenzy.
