AWS Hits 37% Growth, Azure Tops $100B, GCP Launches Flexible Savings Plan: What Q2 2026 Earnings Mean for Enterprise Cloud Contracts

AWS, Azure, and Google Cloud all reported accelerating growth in Q2 2026, with AWS up 37%, Azure up 43%, and Google Cloud up 82%. For enterprises managing cloud commitments, the shared signal is this: capacity is tightening, backlog is surging, and provider leverage in renewal negotiations is increasing across all three. 

AWS: Fastest Growth in 18 Quarters, and the Capacity Story Is Changing 

Here are the results at a glance: 

• Amazon reported Q2 2026 revenue of $200.6B, up 20% year over year 

• AWS revenue reported at $42.2B, up 37%, its fastest growth rate in 18 quarters and well ahead of the 31% analysts had modeled 

• AWS operating income reached $16.6B at a 39.4% margin 

• AWS now accounts for roughly 61% of Amazon’s total operating profit 

• Amazon also raised its full-year 2026 capex guidance to approximately $220B, up from the $200B guided earlier in the year 

• Executives noted that 2027 data center capacity is already largely reserved, with some capacity spoken for as far out as 2028 

That last point is the one clients should sit with. AWS has spent much of the past two years talking about capacity constraints easing. This quarter, leadership effectively told the market the opposite: demand has outrun supply again, and the provider is locking in customer capacity years in advance.  

For clients with an active or upcoming AWS commitment, this changes the negotiating dynamic. AWS has less incentive to offer flexible, ramp-friendly terms when it can allocate scarce capacity to customers willing to commit early and commit big. Clients should expect account teams to push for larger upfront commitments tied to reserved capacity and should scrutinize any capacity-guarantee language carefully before agreeing to a multi-year ramp that assumes AWS will have supply available on your timeline rather than theirs. 

Microsoft Azure: Growth Reaccelerates to 43%, but So Does the Capex Bill 

Here are the results at a glance: 

• Microsoft closed out fiscal 2026 (quarter end June 30) with Azure and other cloud services revenue growing 43% in constant currency, up from 40% the prior quarter and ahead of the roughly 40% Wall Street had expected 

• Azure surpassed $100B in annual revenue for the first time 

• Commercial bookings remain enormous, with commercial RPO (remaining performance obligation) reaching $678B 

• Capex for the quarter came in at $41B 

• While Microsoft’s calendar 2026 capex outlook held at approximately $175B, the company disclosed FY2027 capex guidance of $255B to $260B, a roughly 35% increase, and guided Azure growth of approximately 45% constant currency for the September quarter 

The reacceleration is real and demand-driven, but the capex trajectory is the part clients need to price into their own planning. Two-thirds of Microsoft’s capex is going toward short-lived hardware (GPUs and CPUs on roughly six-year useful lives), and that spend has to be monetized somehow.  

Microsoft has historically used Azure Consumption Commitments and enterprise agreement bundling to lock in growth, and a $678B backlog tells you the company is already succeeding at pulling forward multi-year commitments from its largest accounts. Clients renewing an Enterprise Agreement or Azure commitment in this environment should expect more aggressive true-up and consumption-commitment structures. Push back on baseline growth assumptions that mirror Microsoft’s own AI-driven guidance rather than your organization’s actual, forecasted workload growth. 

Google Cloud: 82% Growth and a Backlog That Signals Where the Leverage Is Going 

Google Cloud’s results were the standout of the three:  

• Alphabet reported Q2 2026 revenue of $119.8B, up 24% year over year 

• Google Cloud revenue surged a massive 82% to $24.8B, comfortably ahead of the 64% growth analysts had projected 

• Cloud operating margin more than tripled year over year, from 20.7% to 35.6% 

• Cloud backlog reached $514B, up more than $50B sequentially from $460B in Q1 

• Alphabet raised full-year 2026 capex guidance to $195B–$205B, up from the $180B–$190B range set just one quarter earlier 

• Management flagged that 2027 capex will increase significantly again.  

• Notably, Google Cloud began recognizing revenue from TPU system sales this quarter, with the majority of currently signed TPU deals expected to convert to revenue in 2027 

An 82% growth rate paired with a backlog that grew by $50B in a single quarter tells clients something important: Google is converting its AI infrastructure position into large, multi-year enterprise agreements at a pace that is starting to close the gap with AWS and Azure. That momentum is good news for clients today, because Google still has more incentive than its larger rivals to win competitive displacement deals and offer aggressive credits to build share. But it’s a closing window.  

As backlog conversion continues and Google’s negotiating position strengthens, the leverage clients have today to extract favorable terms, migration credits, and flexible commitment structures will erode. Clients evaluating GCP as part of a multi-cloud or competitive strategy should treat this quarter as a signal to move now rather than wait. 

UpperEdge’s Perspective: What This Quarter Means for Strategic IT Sourcing 

Every hyperscaler is telling the same story with different numbers: demand is outrunning supply, capex guidance keeps ratcheting upward, and providers are converting that dynamic into larger, longer, and more structured customer commitments. For clients managing existing Monetary Commitment Agreements or heading into a renewal, a few takeaways stand out: 

Capacity scarcity is shifting leverage back to the providers.  

AWS’s comments on 2027–2028 capacity reservations and Azure’s persistent capacity constraints mean clients negotiating new capacity, especially AI/GPU-related capacity, should expect providers to ask for larger, earlier commitments in exchange for supply guarantees. Before your next renewal, plan to get any capacity-guarantee language in writing and tied to enforceable SLAs, not just verbal assurances from the account team. 

Backlog growth is a preview of your next renewal, not just an investor metric.  

Azure’s $678B RPO and Google Cloud’s $514B backlog reflect real commitments other enterprises are already signing. Providers will use those figures internally to benchmark what “reasonable” commitment growth looks like for your account. Don’t let a provider’s aggregate backlog growth become the anchor for your organization’s specific consumption forecast. Understand your consumption trends now to inform your larger negotiation strategy.  

AI is the commercial lever, whether or not AI is your priority.  

All three providers are attributing a meaningful share of cloud growth to AI workloads, and all three are using AI positioning to justify bigger asks in renewal conversations. If AI isn’t yet a significant driver of your own consumption, say so explicitly in negotiations, and resist contract structures that assume AI-driven growth rates you haven’t validated internally. Sales teams for all three providers will be anchoring concessions to AI spend so don’t expect this to be easy! Working now to create internal and executive alignment on messaging here is critical. 

Rising capex doesn’t have to mean rising bills, if you negotiate for it.  

Providers need utilization to justify record infrastructure spend, which creates real incentive on their side to offer credits, discounts, and flexible terms, particularly for competitive displacement or platform consolidation deals. The opportunity exists, but it has to be negotiated for. It won’t show up automatically in your renewal paperwork, so plan to dive deep into your utilization before your next negotiation. 

Google’s new Flexible Savings Plans are purpose-built for AI’s spiky spend pattern.  

Google recently introduced Flexible Savings Plans (FSPs), a spend-based committed use discount that uses a monthly, rather than hourly, entitlement window, initially scoped to Gemini Enterprise and generative AI SKUs. That structure is well suited to AI consumption, which tends to spike and dip in ways traditional hourly CUDs penalize.  

It also comes with the same trade-off as any committed spend model: the commitment is non-cancellable for its full term, and any shortfall in a given month is billed in full with no rollover. Clients piloting or scaling AI workloads on Google Cloud should starting refining their negotiation strategy now to treat FSPs as a tool for locking in a discount on AI spend they’re already confident they’ll consume, not as a way to test AI adoption. Customers should model worst-case underutilization before committing. 

The bottom line for cloud clients:

growth this strong is good news for provider financial health, but it is not automatically good news for your next Enterprise Commitment Agreement. Disciplined internal forecasting, a clear-eyed read on your actual AI adoption trajectory, and a willingness to create competitive tension across providers remain the best defenses against being commercially swept along by a market that is, for now, moving entirely in the hyperscalers’ favor. 

UpperEdge helps enterprises bring clarity, leverage, and control to their cloud negotiations across AWS, Microsoft Azure, and Google Cloud. If your organization has an upcoming renewal or an active Monetary Commitment Agreement, our Cloud Commercial Advisory Services can help you benchmark your terms against what the market is actually seeing.