The AI Capex Reckoning: Oracle Crashed, Chip Stocks Got Dumped, and Software Soared | Week Ending 27 June 2026

The Week in Numbers

The Oracle Implosion: When the AI Bill Arrives

The headline event of the week was Oracle’s collapse — its worst week since the 2001 dot-com bust.

Let that land. Worst week in 25 years.

Oracle closed Friday at $148.53. Its 52-week high was $345.72. The stock has shed more than half its peak value as the market reassesses its AI infrastructure bet. Its 52-week low sits at $134.57 — not far from where it’s trading now.

The numbers tell the story: Oracle’s free cash flow (TTM) is -$24.54B. Negative. The company is spending aggressively on AI data centre build-out, and the cash register is running backwards. With a total debt-to-equity ratio of 388.87%, Oracle is leveraging up heavily to fund this push.

This is not a failing company operationally. Revenue (TTM) is $67.36B, with Q4 FY26 revenue of $19.18B and a profit margin of 25.37%. The P/E (TTM) sits at 25.52, with a forward P/E of 18.87. The average analyst 12-month price target is $252.64 — implying substantial recovery if the AI thesis plays out. YTD the stock is still up 23.33%, which tells you how far it had run before this week’s reckoning.

The structural problem is straightforward: AI infrastructure is expensive. Data centres require land, power, cooling, and hardware — all capital-intensive and all upfront. Revenue from that infrastructure accrues over years. That mismatch between capital outlay and revenue recognition is what the market is now repricing across the entire AI infrastructure layer. The key open question is timing: when does the revenue ramp from these new data centre investments begin to meaningfully offset the heavy upfront capex and return free cash flow to positive territory?

Oracle’s collapse wasn’t just a company-specific story. It was a stress test of the whole AI infrastructure thesis — and this week, it failed.

The Great Chip Divide: Storage Crashes, Logic Holds

The chip selloff quickly separated into winners and losers. The dividing line: are you in the infrastructure supply chain, or the logic and design layer?

Storage and memory got obliterated. Western Digital (WDC) fell 13.17% for the week. Seagate (STX) dropped 12.24%. SanDisk (SNDK) was down 10.45%. Micron (MU) fell 7.47%. These are companies whose revenue is directly tied to data centre build-out volume — if the market questions Oracle-style AI capex, the ripple hits storage immediately.

Analog semiconductors weren’t spared. Texas Instruments (TXN) fell 8.46%, Analog Devices (ADI) dropped 7.42%. Equipment suppliers Applied Materials (AMAT) and Lam Research (LRCX) fell 6.61% and 5.70% respectively. Qualcomm (QCOM) shed 7.57%, Intel (INTC) dropped 3.87% for the week.

Nvidia fell 2.05% for the week — relatively contained compared to peers. Even Nvidia couldn’t fully escape the sector re-rating, though its relative outperformance versus storage and equipment names still reflects the market’s continued confidence in its structural AI position and pricing power.

Citi added fuel mid-week, publishing a note cutting its tech stock weighting with the observation that it’s “difficult to see how everyone in AI/Tech path wins.” That kind of institutional call, arriving at Q2-end during quarter-end rebalancing, accelerates rotation. Fund managers use these moments to reduce positions and lock in gains — and tech had run hard this year.

The message from chip markets this week: if big AI spenders pump the brakes, the entire semiconductor supply chain feels it. Storage is most exposed because it’s volume-driven and commoditised. Equipment suppliers are next. Logic design — with more pricing power — has more cushion.

Why Software Is Winning the AI Rotation

While the hardware layer was getting repriced, something else was happening at the other end of the AI stack. Software stocks surged.

ServiceNow (NOW) led the week, up 10.34%. Microsoft (MSFT) gained 5.71%. IBM matched that. Salesforce (CRM) rose 5.45%. Palantir (PLTR) was up 5.36%. Palo Alto Networks (PANW) added 3.77%.

Microsoft gained 5.71% even while continuing significant Azure capex — a reminder that diversified revenue and strong end-customer monetisation can still win favour when the market scrutinises pure infrastructure bets.

The logic is worth understanding, because it will outlast this week’s headlines.

Software companies don’t build data centres. They use them. They monetise AI by embedding it into workflows that customers already pay for — raising prices, reducing churn, or increasing contract value. The capex sits on someone else’s balance sheet. The software company clips the toll.

This Wednesday’s article on what makes a business “moaty” covers this framework in depth. The market this week delivered a live case study. Capital-light businesses with strong network effects and high switching costs — ServiceNow, Salesforce, Microsoft — demonstrated exactly why those characteristics matter when capital cycles turn. They don’t need to spend $24B building data centres. They need to write code.Re

The same principle appears in other capital-light network businesses. A company like Visa earns a fee on every transaction that flows through its network without owning the underlying money, the banks, or the physical infrastructure. The model is structurally similar to software-layer AI plays — and structurally superior in a rising-rate, capex-scrutiny environment. It shows how durable competitive advantages can exist without heavy capital investment.

The market is repricing “moats.” That’s the week’s real lesson.

The Macro Picture: Sentiment Rebounds, Trade Deficit Blows Out

Underneath the tech drama, the macro data this week was a study in contradictions.

Consumer sentiment surprised strongly to the upside. The University of Michigan Consumer Sentiment reading for June came in at 49.5, beating the consensus of 48.9 and bouncing significantly from May’s 44.8. Consumer Expectations jumped to 50.7 from 44.1. These are substantial moves. Inflation expectations also cooled: 1-year expectations fell to 4.6% from 4.8%, and 5-year expectations dipped to 3.3% from 3.4%.

That’s the good news. The bad news: the US Goods Trade Balance for May blew out to -$105.8B against a consensus of -$85.0B and a prior reading of -$83.01B — the widest deficit since tariff distortions began. The culprit is almost certainly tariff front-loading: importers pulling forward purchases before new levies bite, inflating the import side of the ledger. Economists will discount this distortion, but it feeds negatively into GDP calculations — and it doesn’t look good as a headline.

Fed-wise, Minneapolis Fed President Neel Kashkari stated this week he expects a rate hike this year. That’s a hawkish signal that cuts against market expectations for easing. Kashkari and Williams both spoke on Friday. Markets have been pricing in cuts; Kashkari is on the other side of that debate.

Trump also threatened 100% tariffs on Europe this week. Markets were largely unmoved — suggesting either tariff fatigue or scepticism that they’ll materialise. Either way, the non-reaction is itself telling.

Geopolitical Flashpoint: US Strikes Iran

The week’s most serious macro wildcard was the US carrying out strikes against Iran, after Trump accused Tehran of violating a ceasefire in the Strait of Hormuz.

The Strait of Hormuz is one of the world’s most critical oil chokepoints. Any disruption there has historically spiked crude prices. Yet crude oil fell 2.70% on Friday to $69.98. The market is either pricing in a contained exchange, or reading demand signals more bearishly than supply threats.

Gold climbed 1.01% to $4,088.40 — the classic safe-haven response to geopolitical stress.

The disconnect between falling oil and rising gold is instructive. It suggests the market sees this as a financial-asset risk event (hence gold), but doesn’t believe physical oil supply will be meaningfully disrupted — at least not yet. These situations can escalate, and the coming week’s developments are worth watching closely.

Three scenarios matter most from here: contained tensions (current market pricing), escalation that disrupts oil supply (inflationary and AUD-supportive), or rapid de-escalation (risk-on relief rally). The coming week’s developments will clarify which path is most likely.

What This Means for Australian Investors

Here’s where it gets local — and it matters.

The AUD/USD position: The Australian dollar closed at 0.6896, near the top of its 52-week range of 0.64 to 0.73. For Aussie investors holding unhedged US equities, a strong AUD reduces the translated value of US holdings when you convert back.

But watch the CFTC data: speculative traders turned significantly more net-short the AUD this week, moving from -4.1K to -13.0K contracts. When specs go short, they’re betting the AUD falls. If they’re right, that actually benefits unhedged Aussie investors — a weaker AUD means your USD-denominated holdings translate to more Australian dollars when you convert back.

The yen situation also deserves a note. USD/JPY hit 161.73 this week — the yen at a 40-year low, with Bank of Japan intervention speculation intensifying. A disorderly yen move has historically triggered cross-market volatility, as Japanese institutions unwind carry trades. This is a tail risk worth monitoring.

The ASX comparison: The ASX 200 finished at 8,764.20 (+0.18% Friday), largely shielded from US tech carnage this week. That’s the diversification case in practice. Aussie investors with direct US tech exposure via ETFs or individual names would have felt this week’s moves far more directly.

Kashkari’s rate hike signal: For Aussie investors, a hawkish Fed matters in two ways. First, higher US rates generally support the USD, which pressures the AUD lower. Second, rate hike fears cap US equity multiples — especially for growth names. If Kashkari’s view gains broader FOMC traction, it changes the calculus for US equities.

The AI portfolio question: This week was a live test of how your AI exposure is structured. Broad tech ETFs or AI-themed funds likely took a hit — even though the sector-level story was mixed. The divergence between hardware and software was sharp. Pure-play infrastructure exposure underperformed significantly; software monetisation plays outperformed.

For Australian investors, this has two practical implications. First, review position sizing and consider whether any concentrated infrastructure exposure (via individual names or thematic ETFs) should be paired with or tilted toward the software and platform layer. The capital-light, high-moat characteristics discussed in recent analysis become especially relevant when capital cycles turn and scrutiny on upfront spending increases. Second, the rotation creates natural rebalancing opportunities ahead of EOFY, particularly in SMSFs or taxable accounts where unrealised losses on infrastructure names can offset gains elsewhere while still maintaining AI exposure. Knowing which side of the AI value chain your holdings sit on matters more now than it did six months ago.

What to Watch Next Week

Next week is significant — here’s the calendar that matters.

Close of H1 2026: Final positioning, rebalancing, and window dressing from fund managers. Expect volatility as large funds square books at the half-year mark.

PCE Inflation: The Fed’s preferred inflation measure drops next week. This is the single most important data point for rate expectations. Hot PCE validates Kashkari’s hawkish stance. A cooler reading keeps the door open for cuts later in 2026.

Non-Farm Payrolls (NFP): Moving to Thursday July 2, with US markets closed Friday July 3 (Independence Day observed). A strong jobs number combined with hot PCE would be a troubling mix for rate-cut hopes.

Q2 2026 earnings season: Kicks off in mid-July. This week’s rotation was partly positioning ahead of earnings. Markets will scrutinise AI revenue monetisation commentary closely — particularly from the software names that surged this week. The proof of concept moment is coming.

The Takeaway

The week ending 27 June 2026 will be remembered as the week the market began seriously repricing the AI infrastructure thesis. Not abandoning it — but demanding better answers on the return timeline.

Oracle’s implosion was the clearest signal: you can build the most impressive AI infrastructure in the world, but if free cash flow is -$24.54B (TTM FCF per company filings / Yahoo Finance) and debt-to-equity is 388.87%, the market will eventually ask for proof the investment pays off. That reckoning arrived this week.

The rotation into software wasn’t a rejection of AI. It was a preference for AI business models that don’t require betting the balance sheet first.

The durable lesson threads: capital-light businesses with structural competitive advantages tend to outperform across cycles because they don’t need the market to be generous to generate returns. They earn regardless.

That lesson is older than AI. This week, the market just remembered it.

Wall St. Down Under | Australia

Subscribe | wallstdownunder.com.au

Disclaimer: Wall St. Down Under is an independent financial newsletter for informational and educational purposes only. Nothing published here constitutes financial advice, a recommendation to buy or sell any security, or a solicitation of any investment decision. Always do your own research and consult a licensed financial adviser before making investment decisions. Australian investors should consider their own financial situation, objectives, and risk tolerance. Past performance is not indicative of future results.