THE MOST EXPENSIVE PART OF AI IS WAITING FOR PROOF
SpaceX’s quarterly spending shows why the AI race is becoming a test of investor patience, not only technological ambition.
A company can report faster revenue growth, beat market expectations and still leave investors deeply uneasy. SpaceX has done exactly that.
Its first quarterly results as a listed company showed revenue of $7.8 billion, up 92% from a year earlier. Yet the figure competing for attention was $18.4 billion in quarterly capital expenditure, much of it tied to AI infrastructure and expansion. The company remained cash-flow negative, while its shares fell as the market tried to decide whether early AI revenue justified the scale of the build-out.
THE PROOF ARRIVES AFTER THE BILL
Artificial intelligence creates a difficult sequencing problem. The infrastructure must exist before companies can discover its full commercial value. Computing capacity, data centres, chips, models and specialised talent are not optional extras added after demand appears. They are part of the product being built.
That makes the investment resemble traditional research and development. Factories had to be constructed before mass production became possible. Railways needed tracks before they could move passengers at scale. The difference is that public markets now watch the expenditure unfold in real time and reprice companies every day.
Gareth Edwards captures the pressure in a simple question:
“How much are these tech companies spending on AI and is it going to be worth it in the end?”
Francis Herd’s reply is less comforting:
“Nobody can answer that question.”
WHEN VISION BECOMES A FINANCIAL ASSET
The uncertainty does not affect every company equally. SpaceX carries something many businesses cannot place on a balance sheet: investor faith in Elon Musk’s record of pursuing difficult ideas long before their economics appeared obvious.
That faith can extend the time a company is given to prove itself. It can also make valuation more dependent on execution. SpaceX’s profitable and fast-growing Starlink operation gives the wider group a powerful engine, but investors are still asking whether that engine can support the scale of spending across AI, rockets and future infrastructure.
This is where ambition becomes fragile. A long-term plan remains compelling while milestones keep arriving. Delays, weaker demand or rising costs can rapidly turn patience into doubt.
SOUTH AFRICA IS WATCHING A DIFFERENT MARKET
The episode also exposes a structural gap for South African investors. The discussion contrasts the AI-heavy movements of American markets with a JSE shaped more visibly by mining, gold and platinum. Francis Herd argues that local investors do not have comparable AI-focused giants to follow through the same rapid cycles of fear and optimism.
That does not mean South Africans are excluded from global technology investment. It does mean the local exchange provides fewer direct ways to track the boom, the fear and the sudden reversals surrounding companies building AI at enormous scale.
The SpaceX number therefore reveals more than corporate spending. It shows how the AI race is redistributing risk across markets. Technology companies are asking investors to fund capacity before certainty, while investors are trying to price a future that has not yet fully arrived.
The decisive question is no longer whether AI will change production. It is which companies can survive the cost of waiting for that change to become profitable.
Catch up on all Number of the Day episodes here: https://www.enca.com/number-day-podcast