FEATURED

The AI Spending Cycle Just Hit Its Most Dangerous Moment

Hey there, bargain hunter. Here is where the AI trade actually stands right now, as of July 5, 2026.

The four largest U.S. hyperscalers — Microsoft, Alphabet, Amazon, and Meta — are collectively on track to spend roughly $725 billion in capital expenditures this year, up 77% from an nexalready historic $410 billion in 2025. Add Oracle’s ~$50 billion commitment and the number climbs past $750 billion. That figure has been climbing all year as sell-side estimates kept revising higher. It is not slowing.

The question investors keep dancing around is not whether AI is real. It is whether the returns will arrive fast enough to justify what is being spent.

Here is where it gets uncomfortable. Operating cash flow across hyperscalers is growing roughly 23% per year. Cash capex is growing roughly 70% per year. On current trends, those two lines cross around Q3 2026 — meaning aggregate free cash flow for the group reaches zero. Oracle has already crossed. Amazon is crossing around now. Once capex exceeds operating cash flow, companies have to fund growth through debt, equity, or drawdowns on cash. That changes the risk equation in ways the stock market has not fully worked through yet.

The Market Is Already Splitting

After Q1 earnings, Alphabet surged while Meta dropped 6% and Microsoft slipped. All four beat revenue and earnings estimates. The market did not care. It drew a sharp line between hyperscalers turning massive AI capex into cloud revenue and those still waiting for returns.

That divergence is the signal. Investors are no longer buying the whole basket. They are grading each company individually on one question: is the capital producing revenue?

Alphabet spent $35.67 billion on capex in Q1 alone, more than doubling year over year, while Google Cloud’s contract backlog jumped to over $460 billion. AWS grew 28% in Q1. Microsoft’s AI business now runs at a $37 billion annualized revenue rate, up 123% year over year. The businesses are working. The question is whether the spending stays ahead of the proof.

Slight tangent, but it matters here: investors are demanding proof that large AI capital outlays will lead to a clear acceleration in revenue or margins. At the same time, hyperscalers are shifting from using operating cash flows to increasingly relying on debt markets to fund higher AI-related capital spending. When the funding source changes, the risk profile changes with it.

What Q2 Earnings Will Actually Decide

Banks report mid-July. Microsoft, Meta, and Alphabet all follow in the final week of July. This earnings round matters more than the last one.

The specific threshold to watch: any major hyperscaler trimming forward capex guidance by more than 10% versus its prior framing. The transmission to the broader tech complex is direct. Broadcom guided Q3 AI semiconductor revenue to $16 billion, up over 200% year over year. That number only holds if Microsoft, Meta, Alphabet, and the rest keep writing the checks. Broadcom has also reaffirmed full-year AI semiconductor revenue guidance of $56 billion for fiscal 2026 and is targeting more than $100 billion in AI chip revenue for fiscal 2027. Those are extraordinary commitments that require the hyperscalers to keep spending at pace.

That is the leverage point nobody is discussing loudly enough. Broadcom’s entire growth story, Nvidia’s supply commitments, the whole semiconductor complex — it all runs through whatever the hyperscalers say about 2027 spending when they report.

The One Stock Worth Watching Right Now

Broadcom (AVGO) sits at the center of this entire spending cycle, and after a sharp pullback from its early-June highs, the valuation picture has gotten more interesting.

Here is what the business actually looks like right now. Broadcom reported fiscal Q2 revenue of $22.19 billion, up 48% year over year. AI semiconductor revenue hit $10.8 billion in Q2, up 143% year over year, above guidance. Free cash flow was $10.3 billion, a 46% margin on revenue. The company has six core custom chip customers including Google, Meta, Anthropic, and OpenAI, with long-term multi-year supply agreements. It recently co-developed a custom AI inference chip called Jalapeño with OpenAI — a signal that new customer relationships are moving from early-stage agreements to actual silicon.

Q3 guidance: $29.4 billion in consolidated revenue, up 84% year over year. AI semiconductor revenue projected at $16 billion for Q3, more than tripling from Q2. Full-year AI chip guidance sits at $56 billion, with management targeting $100 billion-plus in fiscal 2027. Those are not incremental estimates. That is a company building toward a fundamentally different revenue base than it had two years ago.

Is It Cheap?

This is where it gets interesting. AVGO is trading around $360-375 (as of early July 2026), well off its all-time closing high of $480.77 reached on June 2, 2026. The 52-week range runs from roughly $262 to $495.

On earnings, AVGO trades at roughly 59.9x — below the semiconductor industry average of about 75.5x and a peer group average of around 81.5x. A DCF-based analysis from Simply Wall St pegs the stock as undervalued by roughly 18.8%, and the company screens as undervalued on 5 of 6 valuation tests using fundamentals-adjusted inputs. Morningstar rates it at a 44% discount to its fair value estimate of $650. Analyst consensus from 48 firms is a Strong Buy with a 12-month price target around $523.

None of that means it is risk-free. Broadcom’s AI revenue is heavily concentrated in a small number of hyperscaler customers. If any major customer trims its custom chip orders, the revenue impact is immediate and visible. The stock dropped roughly 15% in early June when CEO Hock Tan declined to raise the full-year AI guidance above $100 billion — even though Q2 AI revenue came in above forecast. The market punished a non-raise. That kind of expectation dynamic is worth respecting.

But here is the counterpoint. The pullback happened because the stock ran hard into Q2 earnings with aggressive expectations already baked in. The business itself did not disappoint on the numbers. Q2 beat. Q3 guidance was above Wall Street’s estimates. The $100 billion-plus fiscal 2027 target was reaffirmed. Customers are signing multi-year contracts. Free cash flow is substantial. This is not a speculative story waiting for revenue to materialize — the revenue is here and growing at triple-digit rates.

• Revenue growth: Q2 up 48% YoY; Q3 guided at 84% YoY
• AI semiconductor revenue: $10.8B in Q2 (143% YoY); $16B guided in Q3 (200%+ YoY)
• Free cash flow: $10.3B in Q2, 46% FCF margin
• FY2026 AI chip guidance: $56 billion reaffirmed
• FY2027 AI chip target: $100 billion-plus
• Customers: Google, Meta, Anthropic, OpenAI, and two additional undisclosed names
• Current P/E: ~59.9x vs. semiconductor peer average of ~75-81x
• Analyst consensus: Strong Buy (48 firms); 12-month target ~$524
• Key risk: Heavy customer concentration in a small number of hyperscalers

The Structural Risk Hiding in Plain Sight

The broader issue is not Broadcom-specific. Capex is expanding far faster than revenues across the entire hyperscaler group, with a meaningful gap between investment and sales growth. The comparison that should not be dismissed: the 2001 telecom excess cycle, when infrastructure scaled ahead of monetization and the resulting overhang took years to work through. Nobody is calling 2026 the same thing. The businesses are more profitable, the cash flows are real, and the demand signals are genuine. But the structure of the problem — spending racing ahead of monetization, funded increasingly by debt — deserves more serious weighting than the market is currently applying.

Bull / Base / Bear

• Bull: Hyperscalers guide H2 capex higher AND show cloud revenue acceleration. Broadcom’s $16B Q3 AI guide gets confirmed and Q4 goes higher. The stock pushes back toward analyst targets in the $500-plus range.
• Base: Guidance stays roughly flat. Alphabet and Amazon hold. Microsoft and Meta remain under pressure. AVGO trades sideways as the market waits for Q3 results to validate the Q3 guide.
• Bear: A major hyperscaler trims forward capex guidance meaningfully. Custom chip orders get deferred. Broadcom’s $16B Q3 AI revenue guide comes under pressure. The stock tests the lower end of its 52-week range.

What Investors Are Missing

The debate is framed wrong. Most investors are asking “will AI capex continue?” The better question is: what happens to companies that built capacity for $900 billion-plus in 2027 spending if the actual number comes in at $700 billion?

Broadcom’s fiscal 2027 target of $100 billion in AI semiconductor revenue is built on the assumption that the hyperscalers keep expanding their custom silicon footprints. If they do, AVGO at current prices looks like a company trading at a discount to a growth story that is already underway. If they pull back, the $100 billion target gets questioned fast.

That is the question worth sitting with heading into earnings season. Whatever the hyperscalers say about 2027 will either validate every valuation built on this cycle — or start a conversation the market has been actively avoiding. For AVGO specifically, the next six weeks will tell you more about whether this pullback was an opportunity or a warning.

The Cheap Investor