NextDCs OpenAI Campus|SpaceX -16% AI Crash Reprices the Picks and Shovels

The Crash That Hit the Wrong Stock

NextDC fell this session as SpaceX recorded its steepest single-day drop since listing, losing 16.4 per cent to close at US$154.60 on Wall Street overnight. That is a US$600 billion erasure in three trading sessions from a company that raised US$75 billion just eleven days ago as the largest IPO in market history. The immediate trigger was an inaugural bond offering — unspecified in size — the first time SpaceX has asked institutional lenders for money rather than selling equity. A company burning US$4.28 billion in a single quarter does not tap the bond market unless the equity raise alone is insufficient to cover what comes next. The bottleneck here is not the rocket business, which is profitable, and not Starlink, which posted US$4.4 billion in operating profit in 2025. The bottleneck is the AI unit, which recorded a US$6.4 billion operating loss on just US$3.2 billion in revenue last year, and has accelerated its burn into 2026. Investors who bought SpaceX as a satellite story are now holding an AI infrastructure story that has not yet found a way to earn. That repricing belongs to SpaceX shareholders — except that Australia's largest independent data centre operator has already committed to building the physical home for the same AI demand SpaceX cannot yet monetise.

NextDC's Capex Rides the Same Engine

NextDC raised its FY 2026 capital expenditure guidance to between $2.7 billion and $3.0 billion, a number that sits on one foundational assumption: AI hyperscaler demand for compute capacity in Australia will absorb that construction at the contracted utilisation rate of 60 per cent growth the company reported in the March quarter. OpenAI is the named foundational customer for NextDC's US$7 billion AI data centre campus in Western Sydney. OpenAI is the same company that has filed an S-1 for a public listing expected in early 2027, and is currently generating approximately US$2 billion in monthly revenue — but spending at a rate that produced a US$38.5 billion loss last year on US$13.1 billion in revenue. When capital flows into trillion-dollar AI companies at these valuations, analysts have argued the investment case for picks-and-shovels operators like NextDC is strengthened by association. Morgans carries a buy rating on NextDC with a price target of $19, implying approximately 36 per cent upside from where the stock was before this session's sell-off. The standing read is that NextDC is insulated from software valuation risk because it owns the physical infrastructure rather than the AI model. But SpaceX's inaugural bond offering punctures that logic in a specific way: the physical infrastructure is the thing that is losing the money. SpaceX's AI segment spent US$7.7 billion in capital expenditure in Q1 2026 alone, the majority of it on data centres originally built for Grok, and those data centres produced US$818 million in revenue against a US$2.5 billion operating loss in the quarter. The physical layer is not the safe harbour. It is where the losses accumulate while the revenue thesis is still being tested.

The Buried Assumption Behind the Bull Case

The Morningstar analyst team has independently valued SpaceX at US$780 billion — roughly 66 per cent below its post-IPO peak valuation of around US$2.4 trillion. Their specific finding is that the xAI acquisition, conducted without arm's-length pricing given Musk's control of both entities, poses "a material threat of value destruction" to the combined company, estimating more than US$80 billion in potential capital destruction from the AI division alone. The Motley Fool Australia's coverage from June 17 argued the opposite: that the SpaceX IPO wave, followed by Anthropic and OpenAI listings later in 2026 and 2027, would "reallocate institutional capital" toward AI infrastructure supply chains including NextDC, as the AI thesis receives its largest-ever public validation. Both conclusions flow from the same event — the SpaceX IPO — and from the same fact: AI demand is enormous and growing. The divergence is in the assumption each article treats as given. The Morningstar team assumes xAI's competitive position against OpenAI, Anthropic, Google and Meta is not established, and the physical infrastructure built to run Grok may not be monetisable at the rate the valuation requires. The Motley Fool assumes that because AI hyperscalers are spending, the physical infrastructure they need will always be demanded regardless of whether any individual AI company's economics close. For NextDC holders, this distinction is what today's session is actually pricing. The company's Western Sydney campus customer, OpenAI, is spending US$2 billion per month in revenue acquisition while posting annual losses above US$38 billion. If OpenAI's IPO prices at close to US$1 trillion as projected, that customer relationship gets permanently reframed upward for NextDC — demand secured, financing visible, valuation anchored. If the OpenAI IPO follows SpaceX's post-debut trajectory, the repricing runs the other direction. The assumption that physical infrastructure is safe while software is risky fails here because the physical demand is itself a function of AI software valuations not yet closed.

The Checkpoint That Decides the NextDC Thesis

Today mayors of 40 cities, including Melbourne, signed a global data centre pact committing to curb energy and water strain from AI infrastructure buildout. Melbourne's lord mayor noted that NextDC's $2 billion digital hub in Fishermans Bend received ministerial approval last month, and that data centres are projected to account for 10 per cent of local power demand by 2030. The pact is not a shutdown of NextDC's pipeline — it is the first organised regulatory signal that data centre buildout will face conditions that developers have not yet modelled. Combined with SpaceX's 16.4 per cent single-day fall and inaugural bond, there are now two new sources of NXT repricing in today's session alone: the AI valuation signal from the US and the regulatory cost signal from Australia's own municipalities. Neither individually breaks the NextDC thesis. What breaks it is a scenario where OpenAI's IPO in early 2027 does not hold its pre-listing valuation, and hyperscaler spending on Australian infrastructure is subsequently repriced. The monitoring variable for NextDC holders is not the next ASX announcement from the company itself. It is OpenAI's quarterly revenue trajectory between now and its expected 2027 IPO filing. If OpenAI sustains or accelerates beyond its current US$2 billion monthly revenue run rate and posts its first operating profit — which Anthropic was tracking toward in Q2 2026 — the foundational customer relationship at the Western Sydney campus holds its value. If the AI model revenue story stalls before the filing, the US$7 billion campus commitment becomes a question before it becomes an asset. Holders of NextDC should watch OpenAI's revenue disclosures, not NextDC's own capex announcements, as the forward signal. Watch-list investors should wait for the trajectory to resolve before treating the Morgans $19 target as a near-term entry thesis. SpaceX priced at $135, peaked near $192, and closed at $154 — 14 per cent above its offer price, but 20 per cent below its high, on a company still burning US$4 billion per quarter. That is the shape of AI infrastructure repricing: real demand, unresolved economics, and a physical layer that absorbs the capital before the revenue proves the thesis.