Google Cloud on Korea Invest Insights

Alphabet's Q2: Cloud +82% Ends the Demand Debate, Negative FCF Starts the Cash Debate

Thu, 23 Jul 2026 11:00:00 +0900

Context: Who Burns All Those Tokens? argued that final demand, the last weak link in the AI CAPEX debate, had started to get real numbers attached to it, and left the verdict to the earnings season around July 30. The first scorecard arrived earlier than scheduled. Alphabet reported Q2 results in the early hours of July 23 Korea time. This piece synthesizes two analysis notes that dissected the same print from different angles into one. The short version: the demand side of the debate is effectively over, and the debate has moved on to cash flow and return on capital.

TL;DR

Cloud revenue came in at $24.8bn, +82% (vs. consensus +64%), marking five straight quarters of accelerating growth: 32%, 34%, 48%, 63%, 82%. Backlog rose from $462bn to $514bn, a net increase of $52bn even as revenue recognition accelerated.
Demand quality matters more. Existing Cloud customers are consuming on average 50%+ above their initial commitments, and Gemini model API throughput hit 22bn tokens per minute, up 37.5% in a single quarter. With supply running short, Alphabet started, from June, paying SpaceX roughly $920M a month for AI compute. It is an unprecedented sight: a hyperscaler turning into a net buyer of compute.
On the flip side of the same print, quarterly FCF turned negative for the first time ever, at -$5.9bn. CAPEX of $44.9bn outran operating cash flow of $39.1bn, and 2026 CAPEX guidance was raised again from $180-190bn to $195-205bn, with a large further increase flagged for 2027.
The judgment splits three ways. The odds of a post-2028 AI demand cliff have fallen further. Alphabet’s FCF turnaround has been pushed out from 2027 into the 2028-2029 window. For memory, volume demand strength was reconfirmed, but the durability of pricing and peak margins remains a separate question.
We hold the 45/35/20 demand-scenario probabilities. But another strand of hard evidence has piled up in support of the 35% upside case, and the early-CAPEX-deceleration premise central to the 20% downside case was weakened by this print. The next scorecards are Microsoft (July 30 KST) and Amazon (July 31).

Key Framing

Cloud +82% and the first-ever negative quarterly FCF are two sides of the same print. Asked whether the demand is real, Alphabet answered with usage that exceeds commitments and with the act of paying a premium to rent someone else’s servers. What arrived sooner than expected, in exchange, was the bill for serving that demand. The market’s scoring criterion has now shifted from whether demand exists to the cash return after depreciation, and for memory investors this print is reinforcement for volume, not a warranty on margin.

1. The Numbers First: What Came Out

Item	Q2 2026 Actual	Comparison Basis	Verdict
Total revenue	$119.8bn, +24%	Consensus ~$116.9bn	Beat
Google Cloud revenue	$24.8bn, +82%	Consensus $22.4bn, +64%	Large beat
Cloud operating income	$8.8bn (35.6% margin)	Prior year $2.8bn (20.7% margin)	Improved
Search revenue	$63.3bn, +17%	Consensus $63.4bn	In line
Total operating income	$40.8bn, +30% (34.0% margin)	Margin below consensus	Mixed
Adjusted EPS	~$2.85	Consensus ~$2.89	Slight miss
Operating cash flow	$39.1bn	CAPEX $44.9bn	Reversed
Quarterly FCF	-$5.9bn	Prior quarter $10.1bn	First negative ever
Cloud backlog	$514bn	Prior quarter $462bn	+$52bn
2026 CAPEX guidance	$195-205bn	Prior $180-190bn	Raised again

[Fact: Alphabet disclosures and earnings call]

The GAAP numbers carry an optical illusion. Reported GAAP EPS was $9.11 and net income was $112.1bn, up 298% year over year, but that includes roughly $99bn (pre-tax; about $6.26 per share after tax) of unrealized gains on private stakes such as Anthropic and SpaceX. Strip out this operationally unrelated item and the underlying EPS was about $2.85, a slight miss versus market expectations. Revenue and Cloud won big, but the quality of earnings was not as strong as the headline number suggests. [Fact: recalculated from disclosures]

2. The Demand Debate: Why We Think It’s Effectively Over

The previous piece flagged final demand as the weak link and cited Codex’s three million in three days as the first hard data point. Alphabet’s print adds four more strands of hard evidence on top of that.

Usage is running ahead of contracts. Gemini model API throughput rose from 16bn to 22bn tokens per minute, up 37.5% in one quarter, and monthly developers passed 9M. Existing Cloud customers are consuming on average 50%+ above their initial commitments, and the overage widened from the prior quarter. New-customer acquisition is running at twice last year’s pace, and Marketplace transaction volume is up sevenfold. [Fact: CEO letter and earnings call] The hypothesis that demand rests solely on a handful of long-term frontier-lab contracts is not compatible with data showing post-contract actual consumption exceeding commitments.

The demand base has broadened. Alongside model-development demand such as large-scale Gemini 4 pretraining, the company laid out five channels at once: enterprise inference in finance, pharma, retail, telecom and manufacturing; BigQuery and security workloads; consumer services such as Search’s AI Mode and the Gemini app; and outright sales of TPU systems supplied directly into customer data centers. About 90% of the Fortune 100 now use Gemini Enterprise, more than 500 Cloud customers process over 1tn tokens a year, and over 2,000 exceed 100bn tokens. [Fact: CEO letter]

Supply is still capping growth. The CFO stated flatly that the company remains in a supply-constrained environment. To fill near-term shortfalls, Alphabet is leasing third-party data center capacity at elevated prices, and management said roughly six months of inefficiency is worth accepting to win large customers. That is the backdrop for the biggest news of the call: from June, Alphabet has been paying SpaceX roughly $920M a month, about $11bn annualized, to rent AI compute. [Fact: earnings call and press] That the most calculating company in the world is paying a premium to rent someone else’s servers is behavioral evidence that runs directly counter to the demand-overstatement hypothesis.

Backlog grew even while being recognized. Cloud revenue rose about 23.8% quarter over quarter, meaning large contracts are converting into revenue quickly, and yet the backlog balance still grew by another $52bn. The current backlog is about 5.2x annualized Q2 Cloud revenue, and the company said more than half will be recognized within 24 months. [Fact: disclosures and earnings call] As visibility through 2027, that is strong evidence. Customer concentration and the year-by-year recognition amounts beyond 2028, however, were not disclosed. [Blocked: backlog detail undisclosed]

The efficiency counter-argument also lost force this quarter. AI Mode’s cost per response has fallen to its lowest level since launch, yet users and query volume grew even faster. AI Mode has 1bn monthly users, and the Gemini app has 950M MAU. The relationship where usage growth outruns the decline in per-unit cost has now been confirmed at Alphabet’s scale too, running in the same direction as the Jevons pattern observed with Codex. [Inference: efficiency-usage relationship]

3. The Cash Debate: The Bill Arrived Sooner Than Expected

On the other side of the demand story, what this print confirmed is the scale and duration of spending.

Q2 CAPEX of $44.9bn, double the year-ago level, exceeded operating cash flow of $39.1bn, and quarterly capital intensity (CAPEX/OCF) came in at about 115%. The arithmetic is heavy too. With H1 CAPEX at roughly $80.6bn, hitting the top of the annual guidance range of $205bn requires an H2 quarterly average of $57.2-62.2bn, 27-38% above Q2. A meaningful FCF recovery in H2 2026 or in 2027 is unlikely. [Inference: own arithmetic]

There was a hint on 2027 as well. The CFO maintained guidance that 2027 CAPEX will rise sharply, and the market consensus sits at about $257bn. The path from $205bn to $257bn works out to roughly +25% growth. That places the company on a deceleration path, from about +76% this year to +25% next year, not yet a re-acceleration scenario toward the $300bn level. [Fact: earnings call and consensus] This distinction matters. On a deceleration path, the skeleton of the FCF-turnaround thesis stays intact; on a re-acceleration path, the skeleton itself collapses.

Running a 2028 sensitivity in advance sharpens the judgment criteria. The starting point is the current trailing-12-month operating cash flow of $185.7bn.

2028 CAPEX Assumption	OCF Needed for FCF = $0	Required OCF CAGR	OCF Needed for FCF = $50bn	Required Growth
$200bn	$200bn	~3.8%	$250bn	~16.0%
$230bn	$230bn	~11.3%	$280bn	~22.8%
$250bn	$250bn	~16.0%	$300bn	~27.1%

[Inference: own sensitivity, not company guidance]

If CAPEX stabilizes around $200bn in 2028, an FCF recovery is not difficult. If it keeps rising to $230-250bn, combined Cloud-and-Search operating cash flow needs to grow more than 20% a year to get back to prior FCF levels. That is not an impossible number if Cloud’s +82% growth and 35.6% margin hold up for a while, but depreciation, power costs, external lease payments, and a sales mix increasingly weighted toward lower-margin TPU hardware all push in the opposite direction.

A shift in the capital-allocation regime was also formalized this quarter. With $242.5bn in cash and marketable securities and $185.7bn in trailing-12-month operating cash flow, Alphabet’s ability to pay is not in question. But in Q2 it raised about $30.5bn in common equity and about $19.1bn in convertible preferred stock and added about $21.1bn in net debt, while buying back no stock at all. Debt rose from about $16bn to about $100bn in twelve months. [Fact: disclosures] A company that once covered both CAPEX and buybacks entirely from operating cash flow has become one that has paused buybacks for the sake of CAPEX and is now tapping debt and equity capital as well. This is not a liquidity-risk signal but a signal that the capital-allocation regime has changed, and the fact that the weight of the funding sits more with shareholders than with the credit market is the comforting part from a credit perspective.

4. Scoring the Three Questions

Scoring the three questions against this print comes out as follows.

Does AI demand hold up beyond 2028? The odds of a cliff have fallen further. Demand is broadening from training into inference, data, security and enterprise workflows; actual consumption is exceeding commitments; usage growth is outrunning efficiency gains; most TPU external-sales revenue is scheduled to be recognized in 2027; and supply is still short enough that CAPEX must rise sharply again in 2027. The more accurate way to read 2028 is not as the point where demand ends, but as the point where new supply capacity comes fully online and goes on trial.

Will Alphabet’s FCF turn around? Yes, but the timing has slipped. The quarterly FCF crossover arrived sooner than originally expected (early 2027), but even an optimistic 2027 operating-cash-flow estimate of $230-240bn is still FCF-negative for the full year against $257bn of CAPEX. A return to positive FCF requires 2028, and even then only if 2028 CAPEX growth slows to a single digit. A strong FCF re-acceleration is a conditional scenario confined to the 2028-2029 window.

Does this translate into sustained memory purchasing? Strongly positive for volume, neutral for price and margin. We unpack this in the next section.

5. Translating to Memory: Reinforcement for Volume, Not a Warranty on Margin

In Q2, about 60% of technical infrastructure CAPEX went to servers and 40% to data centers and networking. Server spend bundles together TPUs, GPUs, CPUs, HBM, server DRAM, SSDs and networking silicon. [Fact: earnings call]

The items supporting HBM demand are clear-cut: the new TPU generation and the rollout of NVIDIA Vera Rubin, large-scale Gemini 4 pretraining, growing enterprise inference volume, the next-generation network fabric tying together a 1M-accelerator network, and TPU system sales to outside customers, most of which will be recognized starting in 2027. What matters is that even where in-house TPUs partly displace NVIDIA GPUs, HBM demand does not disappear, it simply shifts purchasing channels from the GPU supply chain to the TPU-system supply chain. Add third-party compute leasing on top of that, and HBM and server DRAM demand now flows through three channels: Alphabet’s own data centers, TPU external sales, and leased compute. [Inference: demand-channel decomposition]

The case for server DRAM and eSSD also got thicker with this print. AI agents don’t run on accelerators alone; CPU servers handle preprocessing, state management, security and orchestration. Alphabet highlighted its Axion CPU alongside growing data and security workloads. The figure of 500 Cloud customers processing over 1tn tokens a year shows that storage demand flowing into logs, checkpoints, search indices and vector databases is not confined to training. [Fact: CEO letter] This confirms, via big-tech earnings, the demand-side backdrop for the server DRAM re-tightening covered in the previous piece (Q3 contract prices forecast +13-18%, lead times of 40 weeks).

Even so, this print alone is not grounds for raising memory companies’ 2028 earnings estimates. Just as equipment purchases are surging, memory supply capacity is also set to grow over 2027-2029. Broader in-house chip adoption weakens NVIDIA’s pricing power at the system level, inference cost per response keeps falling fast, and once Alphabet starts tightening CAPEX efficiency, downward price pressure on components will flow down the supply chain. The high prices of HBM4 and next-generation HBM embed early-generation scarcity and a yield premium that are not permanent values. Putting it together looks like this.

Item	Pre-Earnings View	Post-Earnings View
AI accelerator / HBM bit demand	High	Very high
Server DRAM / eSSD demand	Medium-high	High
Advanced packaging / networking	High	Very high
2028 demand cliff	Low probability	Lower probability
Memory ASP durability	Uncertain	Uncertain
Memory peak-margin durability	Low	Low
Early big-tech FCF recovery	Medium	Lower
Incremental return on CAPEX	Unproven	Partially proven, verification ongoing

[Inference: overall judgment]

The 2028 risk for semiconductors is not order cancellations but the normalization of ASP and margin. That is consistent with the 45/35/20 demand-scenario probabilities. This print added one more piece of hard evidence to the 35% upside case and weakened the early-big-tech-CAPEX-deceleration premise behind the 20% downside case. We will update the probabilities themselves once Microsoft and Amazon are also confirmed. From the perspective of Samsung Electronics and SK Hynix, this is a print that improves the demand backdrop heading into the 2027-vintage HBM pricing negotiations that begin in Q4.

6. The Stock Reaction, and Where the Two Notes Diverge

The market reaction summarizes the character of this moment. In after-hours trading, the stock opened down about 2%, briefly recovered to a modest gain, then reversed again once CAPEX guidance was raised on the call, with the decline widening to a 3-5% range. As of the morning of the 23rd Korea time, the share price sat around $342, about 1.5% below the prior close of roughly $347. [Fact: market data] In a market where beating estimates has become the baseline, the 2026 rule that CAPEX decides the reaction played out for a fifth time. Read the other way, the fact that the decline stayed this contained in the face of +82% growth is also a signal that the market will digest the spending as long as the growth keeps proving itself.

The two notes synthesized here read the same facts and reached different conclusions. One applied a structural-proof standard and chose to stay on hold. The accurate phrasing is not that return on capital has been proven but that it is now most likely to be provable, and it withheld a price target given negative FCF, the capital raises, and the flagged 2027 increase. The other maintained a buy on an entry-price basis. Taking a price already 15% off its peak as the starting point, it applied a 28-29x multiple to about $15 of ex-gains 2027E EPS to arrive at a 12-month target of $430, with a withdrawal condition attached: it will be recalculated if 2027 CAPEX is quantified near $300bn. [Fact: the two notes’ conclusions]

The difference between the two judgments is not a difference in facts observed but a difference in time horizon and standard. The structural view puts the burden of proof on the company until the cash return after depreciation is confirmed in the numbers; the price view calculates how much of that risk is already reflected in a stock that has already corrected. In keeping with this blog’s principle of not recommending a specific direction, we present both frames side by side, but the common denominator is clear. Either way, the next variable to be judged is not demand but cash.

7. Scorecard: The Next Two Weeks Will Decide

Here is a rundown of the boxes this print filled in, and the boxes it newly opened.

Alphabet’s share of the big-tech CAPEX commentary is now confirmed as a raise. What remains is Microsoft’s first FY27 guidance (call in the early hours of July 30 KST) and Amazon’s AWS growth rate (early hours of July 31). If Alphabet closed out the demand debate, these two will settle the margin and FCF debate.
Whether Cloud growth holds above 50% for the next 2-4 quarters and whether margin holds above 30% are the confirming indicators for the upside case. If growth falls below 30% once supply constraints ease, the possibility of demand overestimation needs to be recalculated.
Watch whether backlog keeps net-adding even after revenue recognition. A combination of a shrinking backlog, or one where only contract duration keeps extending, would signal deteriorating contract quality.
The quarter in which trailing-12-month operating-cash-flow growth overtakes CAPEX growth is the starting point of the FCF turnaround. Conversely, if 2028 CAPEX exceeds $250bn while operating cash flow falls short of that pace, we will call it structural FCF impairment.
Whether large additional equity or debt raises continue, whether SpaceX-style external leasing shrinks back in 2027-2028, and whether TPU external sales convert into recurring Cloud consumption are the checkpoints on the capital-efficiency side.
The discriminator on the memory side is unchanged: DRAM contract pricing, and confirmation of Q3 contract prices on the Samsung Electronics and SK Hynix calls around July 30.

Names mentioned in this piece are examples for analysis and are not a recommendation to buy or sell any specific security. Responsibility for investment decisions and their outcomes rests with the investor. Alphabet did not provide 2028 CAPEX or Cloud revenue guidance; the customer mix, year-by-year recognition amounts and cancellation terms behind the $514bn backlog were not disclosed; and purchase volumes and supplier mix for HBM, DRAM and NAND are also not disclosed items. The 2028 FCF sensitivity and the H2 CAPEX arithmetic are our own estimates starting from current disclosures, not company guidance. The pre-tax size of the unrealized gain on private-equity stakes and its after-tax per-share effect are recalculations based on disclosures and may differ slightly due to rounding. The $430 target is the calculation of one of the notes synthesized here, not this blog’s target price. Share prices and quotes are as of the morning of July 23, 2026 Korea time.

Big Tech AI Earnings: 2027 Demand Is Contracted, but 2028 Free Cash Flow Is the Real Test

Wed, 22 Jul 2026 17:30:00 +0900

Amazon, Microsoft, and Alphabet plan to spend roughly $570-580 billion on capital investment in 2026. Their latest reported cloud growth rates were 28%, 39% in constant currency, and 63%, respectively. The market is no longer debating whether current AI demand is strong. It is asking whether the data centers built now will be well utilized, whether revenue can outrun depreciation and power costs, and whether memory profits can survive the supply response expected from 2028.

The last four quarters suggest that much of the demand through 2027 is already supported by contracts, reservations, and supply constraints. Beyond 2028, scarcity is no longer enough. Utilization, inference revenue, customer concentration, and cash collection become the decisive variables.

Related context This is a follow-up to the July Big Tech earnings and memory thesis preview. The financing side is covered in Hyperscaler financing and the memory bottleneck, while the semiconductor supply response is developed in the weekly semiconductor deep dive.

Summary

AI demand is likely to remain stronger than expected through 2027. Backlogs, long-term commitments, supply shortages, and customer expansion point in the same direction.
The most durable demand is likely to come from enterprise inference, agents, coding, search, and data services rather than one-off frontier training runs.
The three companies plan approximately $570-580 billion of 2026 capital spending. Spending can grow again in 2027, although the growth rate should slow from an unusually high base.
Return on investment should be measured through incremental cloud gross profit and first-party revenue or cost savings, less depreciation, power, and financing costs.
Current evidence supports the highest probability of post-2028 free cash flow recovery at Alphabet, followed by Microsoft and Amazon. Alphabet shows the clearest combination of monetization and unit-cost improvement. Microsoft has the strongest contracted visibility but more concentration and lease complexity. Amazon has the largest rebound option but the greatest current cash-flow pressure.
Memory purchases should continue across HBM, server DRAM, and enterprise SSDs. Custom accelerators change the product mix but do not remove the need for memory.
The principal risk after 2028 is price rather than bit demand. New capacity can reduce average selling prices and margins even if memory shipments keep rising.
Credit conditions are cautionary, not distressed. Bond books remain covered, but lower cover ratios and wider long-end spreads show that capital markets are beginning to impose discipline.

Core judgment

Through 2027, supply constraints and contracted demand support continued spending. From 2028, the decisive variables shift to utilization, cash collection, and returns after depreciation. For memory, strong volume and normalizing prices can coexist.

1. The three earnings calls and the questions that matter

Alphabet reports at 5:30 a.m. KST on July 23, Microsoft at 6:30 a.m. KST on July 30, and Amazon at 6:00 a.m. KST on July 31, based on company investor-relations announcements.

Company	Earnings time	Primary question
Alphabet	July 23, 05:30 KST	How quickly do 63% cloud growth and a $462.3 billion backlog convert into revenue and cash?
Microsoft	July 30, 06:30 KST	When does Azure supply improve, and does backlog growth reaccelerate outside OpenAI?
Amazon	July 31, 06:00 KST	Can AWS protect margins and restore free cash flow while Amazon spends about $200 billion?

The market is trying to answer three questions. How long can AI demand grow above expectations, and from which customers and workloads? How much longer can capital spending rise, and can the investments earn adequate returns? Will that spending translate into recurring purchases of HBM, server DRAM, and enterprise SSDs?

The last four quarters provide a constructive answer to the first question. The second and third require company-specific analysis because a dollar spent on accelerators has a different return profile from a dollar spent on land, power, networking, or long-lived buildings.

2. Four quarters of operating evidence

2.1 Amazon: AWS accelerated while cash left first

Quarter	AWS revenue	YoY growth	AWS operating income	Cash-flow signal
Q2 2025	$30.9B	+17.5%	$10.2B	TTM FCF $18.2B
Q3 2025	$33.0B	+20%	$11.4B	AI infrastructure spending kept rising
Q4 2025	$35.6B	+24%	$12.4B	Amazon signaled about $200B of 2026 investment
Q1 2026	$37.6B	+28%	$14.2B	TTM FCF fell to $1.2B

AWS growth accelerated for four consecutive quarters. Its Q1 operating margin reached 37.8%, showing that revenue and profit expanded despite supply constraints.

The cash profile moved in the opposite direction. Amazon generated $148.5 billion of trailing operating cash flow, but net property and equipment spending reached $147.3 billion. AWS-centered long-term commitments stood at $364 billion with a weighted-average remaining life of 5.5 years. Revenue visibility is high, but spending occurs before the associated revenue arrives.

Amazon’s 2025 shareholder letter stated that the company was not spending $200 billion on a hunch and had substantial customer commitments. Graviton, Trainium, and Nitro together exceeded a $20 billion annual revenue run rate and were growing at a triple-digit rate. Custom silicon is both a cost-control tool and a way to keep workloads inside AWS.

Amazon can produce the largest cash-flow rebound if the new assets fill quickly, but it also has the widest gap to close. AWS needs to maintain growth around the high-20% range and defend margins in the mid-to-high 30s while 2026 investments begin monetizing in 2027 and 2028.

Sources: Amazon Q1 2026 results, Amazon 2025 shareholder letter, Amazon Q1 2026 10-Q

2.2 Microsoft: the strongest contract visibility and the largest concentration question

Quarter	Revenue	Azure growth, constant currency	RPO	Capital spending	Free cash flow
FY25 Q4	$76.4B	about +39%	$368B	$24.2B	$25.6B
FY26 Q1	$77.7B	+39%	$392B	$34.9B	$25.7B
FY26 Q2	$81.3B	+38%	$625B	$37.5B	$5.9B
FY26 Q3	$82.9B	+39%	$627B	$31.9B	$15.8B

Azure grew 40% as reported and 39% in constant currency in the latest quarter. Microsoft’s AI business exceeded a $37 billion annual revenue run rate and grew 123%. Commercial RPO reached $627 billion with an average duration of 2.5 years, and roughly 25% is expected to convert to revenue over the next 12 months.

Contract quality needs to be separated from contract size. OpenAI accounted for approximately 45% of commercial RPO in FY26 Q2, while RPO excluding OpenAI grew 26% in FY26 Q3. A major strategic customer and a broader enterprise expansion are both present.

Microsoft also has the most complex capital-spending definition. Q3 cash property and equipment spending was $30.9 billion, while spending including finance leases reached $31.9 billion. Management said roughly two-thirds of capital spending went to shorter-lived assets such as GPUs and CPUs, with one-third allocated to land and buildings. Short-lived assets monetize earlier but also return to the replacement cycle sooner.

Azure is expected to remain supply constrained through 2026, with modest growth acceleration possible in the second half of the calendar year. Q4 capital spending is expected to exceed $40 billion. Calendar 2026 investment is approximately $190 billion, including about $25 billion of component inflation. Microsoft expects double-digit revenue and operating-income growth in FY27.

The company has the strongest contracted visibility, but post-2028 free cash flow depends on independent enterprise demand, Copilot and agent usage, lease-adjusted cash generation, and whether gross profit grows faster than depreciation.

Sources: Microsoft FY26 Q3 earnings call, Microsoft FY26 Q3 results

2.3 Alphabet: demand, margin, and unit-cost improvement are visible together

Quarter	Google Cloud revenue	YoY growth	Operating income	Margin	Backlog	Capital spending
Q2 2025	$13.6B	+32%	$2.8B	20.7%	$106B	$22.4B
Q3 2025	$15.2B	+34%	$3.6B	23.7%	$155B	$24.0B
Q4 2025	$17.7B	+48%	$5.3B	30.1%	$240B	$27.9B
Q1 2026	$20.0B	+63%	$6.6B	32.9%	$462.3B	$35.7B

Alphabet delivered the clearest simultaneous improvement. Cloud growth accelerated from 32% to 63%, the operating margin rose from 20.7% to 32.9%, and backlog increased from $106 billion to $462.3 billion.

Q1 generative AI product revenue grew 800%. New customers doubled, deals between $100 million and $1 billion doubled, and Alphabet signed multiple deals above $1 billion. Existing customers consumed 45% more than their original commitments. Gemini Enterprise paid monthly users rose 40% quarter over quarter.

Unit economics also improved. Alphabet said the cost per AI response fell by more than 30%. Higher demand and lower unit cost are the best combination for infrastructure returns. The same infrastructure supports Search, YouTube, Workspace, and advertising, so external cloud revenue understates part of the return.

Alphabet guides to $180-190 billion of 2026 capital spending and expects another significant increase in 2027. Q1 uncommenced asset commitments were approximately $108.6 billion, and uncommenced data-center leases were $75.6 billion, scheduled to begin between 2026 and 2031. Total RPO was $467.6 billion, with just over half expected to be recognized within 24 months.

The main risk is the speed of spending itself. Q1 operating cash flow was $45.8 billion and capital spending was $35.7 billion, leaving quarterly free cash flow of $10.1 billion. Trailing free cash flow remained a substantial $64.4 billion, but another large increase in 2027 spending could delay the recovery.

Current evidence nevertheless places Alphabet first in proving AI investment returns. The next call needs to clarify how much backlog growth came from a few large contracts and how quickly new facilities are filling.

Sources: Alphabet Q1 2026 earnings call, Alphabet Q1 2026 10-Q

3. Where AI demand can persist

AI demand should be divided into five layers with different customers and durations.

Demand layer	Customers	Expected durability	Key evidence
Frontier training and post-training	Leading model labs and first-party models	Strong but concentrated	Cluster reservations, long-term power contracts, model roadmaps
Enterprise inference and agents	Finance, retail, manufacturing, healthcare, government	Most durable candidate	Paid users, calls, spend per customer, renewals
Development and coding	Software companies and developers	Rapid adoption, strong price competition	Seats, agent runtime, generated code, retention
Data, storage, and general cloud	Existing cloud customers	Stable and bundled with AI	Database, storage, networking, migrations
First-party services	Search, advertising, productivity, commerce, recommendations	Revenue and cost savings combined	Conversion, cost per response, productivity, gross margin

3.1 Supply should remain the constraint through 2027

AWS, Azure, and Google Cloud have all described supply constraints. Alphabet customers used 45% more than their initial commitments. Microsoft expects constraints through the end of 2026. Amazon’s long-term commitments have an average 5.5 years remaining.

This supports a capacity-led revenue model through at least 2027. As data centers open and power is connected, revenue should follow. Near-term spending reductions are also difficult because construction, leases, power, servers, and memory are already under contract.

3.2 2028 is more likely to be a change in demand quality than a demand cliff

More facilities and semiconductor supply arrive from 2028. Once scarcity eases, investors must test whether customers actually use reserved capacity, renew contracts, and increase usage fast enough to offset lower inference prices.

Demand can remain high while profitability falls. Model efficiency and custom silicon reduce cost per task, while cloud competition can lower revenue per token. If call volume grows faster than pricing declines, utilization and revenue still rise.

The useful relationship is:

AI revenue growth = workload growth + customer growth - price decline per workload

Enterprise agents, coding, and embedded inference have the best chance of producing high-frequency, recurring use.

3.3 Independent customers are the most important proof

Strategic customer contracts support early demand, but reciprocal investment and revenue commitments can overstate final demand. Confidence after 2028 requires more usage from enterprises, governments, developers, and smaller customers without financing links to the infrastructure provider.

Alphabet’s doubled new-customer count, doubled large-deal count, and commitment overages are the clearest current evidence. Microsoft’s 26% RPO growth excluding OpenAI is more important than total RPO alone. Amazon needs to disclose more about how much of AWS commitments comes from AI versus general cloud demand.

4. How much longer can capital spending grow?

4.1 Approximately $570-580 billion in 2026

Company	2026 investment plan	Definition caveat
Amazon	about $200B	Primarily AI and AWS, with some spending elsewhere
Microsoft	about $190B	Cash spending and finance leases can both be included
Alphabet	$180-190B	Servers, data centers, networks, and custom silicon
Total	about $570-580B	Accounting definitions differ across companies

The absolute amount can grow again, but the 60-100% growth rates seen in 2026 are difficult to repeat. The base is larger, while power, sites, transformers, networking equipment, and skilled labor remain constrained.

The following 2027 ranges are analytical scenarios, not company guidance.

Scenario	Combined 2027 spending	Conditions
Downside	$530-580B	Slower bookings, power delays, digestion of 2026 prebuys
Base	$630-665B	Continued constraints, slower growth rate, execution of existing commitments
Upside	Above $700B	Agent usage surge, additional sovereign demand, custom silicon and GPU expansion

A slower capital-spending growth rate is not by itself a bearish signal. The warning comes when slower spending, cloud growth, and backlog appear together.

4.2 A simple return framework

AI investment return = incremental cloud gross profit + first-party revenue and savings - depreciation - power - financing cost

Capital spending hits the cash-flow statement first. Depreciation enters the income statement after servers and data centers begin operating. Cash-flow troughs and accounting-margin troughs therefore occur at different times.

Free cash flow is pressured first in 2026. Revenue and depreciation rise together in 2027-2029. Around 2030, accelerators purchased in 2026 can begin returning to the replacement cycle. Cash flow can recover before reported margins fully absorb depreciation.

4.3 Current evidence by company

Company	Positive evidence	Remaining risk	Confidence in post-2028 FCF recovery
Alphabet	Cloud +63%, margin 32.9%, cost per response down more than 30%	Fast spending growth, leases, possible large-deal concentration	Highest
Microsoft	Azure +39%, RPO $627B, AI ARR above $37B	OpenAI concentration, finance leases, short-lived assets	Medium-high
Amazon	AWS +28%, margin 37.8%, commitments $364B	TTM FCF $1.2B, $200B spending, monetization lag	Medium

5. The dashboard for 2028 and beyond

Metric	Constructive signal	Warning signal
Independent-customer RPO	Growth outside strategically financed customers	A few customers drive most of the increase
Next-12-month recognition	A larger share of RPO converts soon	Contract duration rises while near-term conversion slows
Cash collection	Cash receipts outgrow contract assets	Receivables and contract assets grow without deposits
Operating cash flow	Four-quarter OCF growth exceeds capex growth by at least 10 points	OCF trails investment growth
Cloud economics	Gross profit growth exceeds depreciation and power growth	Revenue grows while margin falls
Utilization	New capacity fills rapidly	Low utilization after power connection
Lease-adjusted FCF	Recovery remains after finance leases	Reported FCF improves while lease obligations rise
Custom silicon	Lower cost and higher customer use	R&D rises without adoption
Memory contracts	Commitments convert to shipments and cash	Deferrals, renegotiation, or cancellations
Supply normalization	Shipments and utilization hold despite lower prices	ASP and utilization fall together
Replacement cycle	First-generation assets earn returns before replacement	Replacement begins before initial payback

The strongest combination is independent RPO growth, new capacity going online, stable cloud margins, and OCF growth overtaking capital-spending growth. The most dangerous combination is easing supply constraints, slower growth, higher depreciation, rising customer concentration, and increasing debt or leases.

6. Credit markets are imposing discipline, not signaling a crisis

Apollo reports that hyperscaler bond cover ratios fell from around 5 times in February 2026 to below 2 times in July. Demand still exceeds supply, but investors have become more selective.

Amazon issued $25 billion of bonds on July 7. Orders peaked near $62 billion and ended around $41 billion after pricing. Its longest maturities offered an additional 18-21 basis points of yield. Funding remained available, but at a higher cost.

The Bank of England’s July 2026 Financial Stability Report said AI-related issuance had not yet impaired broader investment-grade terms or availability. Barclays projected approximately $240 billion of 2026 investment-grade hyperscaler financing, comparable to major bank issuance.

The signal is yellow, not red. Capital markets remain open but increasingly demand backlog quality, utilization, and cash conversion. This should delay weaker projects and widen capital-cost differences rather than end AI spending outright.

Sources: Apollo hyperscaler bond cover ratios, Bank of England July 2026 Financial Stability Report, Amazon bond demand

7. Does the spending become recurring memory demand?

7.1 AI systems require a memory hierarchy

Memory layer	Function	Demand direction
HBM	Bandwidth close to GPUs and ASICs	Strongest in HBM3E and HBM4
Server DRAM	CPU work, agent orchestration, preprocessing, caches, metadata	Expands with agents and multi-model systems
Enterprise SSD and NAND	Checkpoints, vector databases, documents, video, long-context storage	Expands with inference volume and retention

Training is HBM intensive. Enterprise inference and agents expand server DRAM and SSD demand alongside HBM. Multiple agents repeat tasks and preserve longer context, increasing memory outside the accelerator.

TPUs, Trainium, and Maia can displace some external GPUs but still require HBM, server DRAM, and storage. Custom silicon changes specifications, packaging, and bargaining power rather than eliminating memory demand.

7.2 Procurement is indirect and arrives with a lag

Hyperscalers do not purchase all HBM directly. Procurement runs through Nvidia, custom ASIC partners, server ODMs, and packaging suppliers. Capital spending therefore reaches memory-vendor revenue with a lag through design, allocation, wafer starts, packaging, server assembly, and installation.

Micron’s strategic customer agreements illustrate the connection. It has 16 agreements, generally covering 2026-2030, representing roughly 20% of DRAM volume and one-third of NAND volume. Fourteen agreements contain approximately $100 billion of cumulative minimum-price revenue, supported by about $22 billion of customer deposits and financing commitments.

Long-term agreements raise the earnings floor. Take-or-pay and minimum prices reduce cancellation and downside pricing. Ceiling prices can also limit upside when spot prices surge. These contracts reduce the severity of a downturn but do not permanently lock in today’s scarcity margins.

Source: Micron FY3Q26 prepared remarks

7.3 The 2028 memory risk is price, not necessarily demand

Additional DRAM, NAND, and HBM packaging capacity should arrive from 2027 and 2028. Bit shipments can keep growing while average selling prices and margins normalize.

Memory profit = bit shipments × price per bit - expansion, depreciation, and input costs

In 2026 and 2027, demand is growing faster than supply, supporting both volume and price. From 2028, supply growth can reduce pricing even if bit demand stays high. Micron’s current gross margin above 80% reflects scarcity and should not be treated as a steady-state margin.

The semiconductor conclusion therefore requires more than a direction for Big Tech capital spending. Utilization, memory supply growth, contract pricing, inventory, and producer capital discipline must be read together.

8. Company-specific earnings checklist

Alphabet

Cloud growth near 60%, with both new customers and usage increasing.
Backlog rising from $462.3 billion without a deterioration in near-term recognition.
Continued explanation for $180-190 billion of 2026 spending and the 2027 step-up.
Cloud margin staying above 30% and cost per response falling further.
Evidence that external TPU sales and internal usage both scale.

Microsoft

Azure growth near 39-40% after additional supply comes online.
RPO growth reaccelerating outside OpenAI.
More than $40 billion of quarterly spending converting into revenue quickly.
Paid Copilot and agent seats accompanied by actual usage and revenue.
Operating cash flow growing faster than capital spending and finance leases.

Amazon

AWS growth in the high 20s and margins in the mid-to-high 30s.
Faster conversion of the $364 billion commitment base.
Trainium adoption and revenue demonstrating a cost advantage over external GPUs.
A clear 2027-2028 monetization schedule for roughly $200 billion of investment.
Operating cash flow rising while net investment growth slows.

9. Three paths for 2028-2031

These probabilities are analytical judgments, not company guidance.

Base case, 55%

Capital spending rises again in 2027 but at a slower rate. New data centers fill, and enterprise inference and agent revenue expand. Alphabet recovers free cash flow first, followed by Microsoft. Amazon recovers later because of a longer monetization lag.

Semiconductor revenue remains high, but new supply normalizes memory prices and margins from 2028. HBM and server DRAM volumes remain strong while explosive spot-price increases fade.

Upside case, 20%

Agent, coding, search, and enterprise inference usage grows faster than unit-price declines. Sovereign and enterprise customers add demand. Combined 2027 capital spending exceeds $700 billion while operating cash flow also expands.

HBM4, advanced packaging, server DRAM, and enterprise SSD shortages persist into 2028. Memory contract volumes and prices rise, producing another round of semiconductor earnings upgrades.

Downside case, 25%

Model efficiency and price competition reduce revenue per task quickly. Strategic contracts are renegotiated, and new data centers operate below expected utilization. Depreciation and power costs are already committed, pressuring margins and free cash flow.

Memory supply added in 2027-2029 exceeds demand growth, pushing down ASP and utilization together. Long-term agreements soften the decline but do not protect uncontracted volume and spot pricing.

10. Conclusion

AI demand is likely to remain stronger than expected through 2027. Cloud growth, backlogs, supply constraints, and long-term infrastructure commitments support that view. The customer base is broadening from frontier labs to enterprise inference, coding, search, and general cloud workloads.

Capital spending should remain elevated in 2027, although its growth rate can slow. A slowdown from a very high base is not a cycle break. The real warning would be simultaneous weakness in backlog, utilization, cloud margins, and operating cash flow.

Current evidence gives Alphabet the best chance of post-2028 free cash flow recovery. It has shown demand growth, margin improvement, and lower unit costs together. Microsoft has stronger contract visibility but more OpenAI concentration and finance-lease complexity. Amazon has powerful AWS economics and significant rebound potential, but it must close the gap between $200 billion of investment and trailing free cash flow of only $1.2 billion.

Memory purchases should continue, but the decisive post-2028 issue is pricing after supply expands. Custom silicon and inference efficiency change the mix across HBM, server DRAM, and enterprise SSDs rather than eliminating demand. Strategic customer agreements raise the floor but do not make current scarcity margins permanent.

The most important figures in these earnings calls are not capital spending totals alone. Independent-customer backlog, next-12-month conversion, utilization, cloud margins, and lease-adjusted free cash flow must move together. If they improve, 2028 becomes the start of cash harvesting. If customer concentration rises while utilization and cash collection lag, today’s earnings strength can return as tomorrow’s depreciation burden.

This article is based on public company investor-relations materials, SEC filings, and industry sources. Backlogs and long-term agreements improve visibility but remain subject to usage, recognition timing, pricing, and contract terms. This is informational analysis, not a recommendation to buy or sell securities.

Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.