Oracle’s $553B AI Backlog Could Stall as Power Bottleneck Looms in 2027–2028

Oracle is now squarely in the early, high-growth phase of the AI infrastructure S-curve. The company's financials show it is not just participating in the boom but is actively building the rails that will carry it forward. The core of this thesis is quantified in its remaining performance obligations (RPO) of $553 billion, which grew a staggering 325% year-over-year last quarter. This backlog isn't speculative-it's a massive, contracted pipeline of future revenue that validates the market's appetite for its AI data center strategy.

This backlog directly translates to Oracle's near-term financial trajectory. The company has used this momentum to raise its revenue forecast for fiscal 2027 to $90 billion, a clear beat against analyst estimates. The setup is straightforward: OracleORCL-- is capturing large-scale AI deals where customers fund the upfront semiconductor purchases, allowing Oracle to scale its cloud infrastructure without immediate capital strain. This is the classic early adopter phase, where securing the backlog is the primary goal, and Oracle has executed masterfully.

The rapid adoption is also visible in its core cloud engine. Oracle's infrastructure revenue surged 84% year-over-year to $4.9 billion last quarter, accelerating from the prior period. This isn't just growth; it's exponential scaling, demonstrating that the demand for its AI compute capacity is outpacing supply. For an investor, this is the signal that Oracle is positioned at the right moment in the technological paradigm shift. It has built the fundamental infrastructure layer for AI, and the backlog proves the market is paying for it. The next phase will be margin expansion as this massive installed base begins to generate profits, but for now, the S-curve is bending sharply upward.

The Infrastructure Bottleneck: Power as the New Compute Constraint

The AI boom is hitting a physical wall. While Oracle's backlog and revenue surge show it is building the compute rails, the entire S-curve faces a fundamental constraint: energy. The projected spending is staggering. Major tech firms have announced plans to spend over $600 billion on AI in 2026 alone. This wave of investment is already fueling unease, not about profitability, but about the sheer physical scale of the build-out. The largest AI data centers consume over a gigawatt of continuous load, enough to power hundreds of thousands of homes. This is the new bottleneck.

The risk is clear. If the energy infrastructure cannot scale to meet this demand, the entire AI compute buildout faces delays. Developers expect power constraints to emerge as early as 2027–2028 due to underinvestment in grids and regulatory red tape. This isn't a distant theoretical problem. It's a direct threat to the demand for Oracle's data center services. If a customer cannot secure power, they cannot deploy the AI infrastructure Oracle is selling them. The exponential growth in compute demand is now inextricably linked to a parallel, and potentially slower, build-out of power generation and transmission.

Oracle's strategy offers some insulation. By partnering with hyperscalers like OpenAI and Meta, it is embedded in the front lines of this expansion. Its model of having customers fund the upfront semiconductor purchases also shifts some capital risk. Yet, Oracle remains dependent on a broader ecosystem now facing a physical bottleneck. The company's own capital expenditures were $18.6 billion last quarter, a massive bet on its own infrastructure. But even Oracle's capacity is ultimately limited by the power grid and the availability of independent energy plants, a trend already visible as data centers plan to build their own power plants.

The bottom line is that the AI S-curve is not just about chips and software. It is a system of interdependent layers, and the energy layer is now the critical constraint. For Oracle, the risk is that its massive backlog and infrastructure revenue growth could stall if the power bottleneck materializes sooner than expected. The company is building the rails, but the track itself may not be ready.

Financial Impact and Valuation: Growth vs. Grid Risk

Oracle's latest quarter delivered the financial proof point the market needed. The company posted adjusted earnings per share of $1.79 on a 22% year-over-year jump in revenue to a record $17.2 billion, crushing analyst estimates. This beat, coupled with a raised long-term outlook, provided a "huge relief" for a sector jittery about the AI buildout, according to analysts. Yet the stock's reaction tells a deeper story of skepticism. Despite the strong results, shares remain more than 50% off their September highs, trading at a price-to-earnings ratio of 26.8. This gap between robust fundamentals and a depressed valuation is the market's bet on a future risk: the power bottleneck.

The core of Oracle's valuation hinges on one exponential conversion: turning its $553 billion backlog into cash flow. The company's raised forecast for fiscal 2027 revenue of $90 billion implies a steep ramp in the conversion rate. But this timeline is now vulnerable. The market is pricing in the risk that project execution could stall. Developers expect power constraints to emerge as early as 2027–2028 due to underinvestment in grids. Any delay in securing energy for these massive AI data center projects would directly pressure the backlog-to-cash conversion, stretching out the path to profitability and justifying the current P/E discount.

The bottom line is a tension between two S-curves. Oracle is riding the steep, early phase of the AI compute demand curve, as evidenced by its soaring revenue and backlog. But the entire paradigm shift faces a physical constraint on the energy supply curve. The stock's valuation reflects this duality: it rewards the massive contracted demand while heavily discounting the risk that the infrastructure to deliver it-power-may not be ready on time. For now, the backlog is the anchor. But the grid is the variable that will determine how fast the anchor can be pulled in.

Catalysts and Risks: The Path to 2027

The investment thesis now hinges on a race between execution and infrastructure. Oracle has set a clear target: convert its massive backlog into the $90 billion in revenue for fiscal 2027. The near-term catalyst is successful delivery. The company must demonstrate it can manage the logistics of these large-scale AI contracts, particularly the capital-intensive build-out of data centers. Announcements of new partnerships or the successful commissioning of projects with partners like OpenAI and Meta will be critical proof points. The market needs to see that the 84% year-over-year surge in infrastructure revenue is not a one-off but the start of a sustained ramp that meets the raised outlook.

The primary risk, however, is external and systemic. The entire AI S-curve faces a physical bottleneck in energy. The market is pricing in a delay, with developers expecting power constraints to emerge as early as 2027–2028. This isn't a minor hiccup; it's a potential cascade. If regulatory red tape or supply-chain shortages in turbines and grid components slow the build-out of power, it will directly delay data center construction timelines. For Oracle, this means managing customer expectations or, in a worst-case scenario, facing contract revisions. The company's own $18.6 billion in capital expenditures last quarter shows its commitment, but it cannot control the broader energy supply chain. The Reuters report highlighting that tech giants have announced plans to spend over $600 billion on AI in 2026 alone underscores the scale of the coming strain on grids.

The key watchpoint is the quarterly data. Investors must monitor two metrics for signs of acceleration or deceleration. First, the growth in $553 billion remaining performance obligations will signal whether the AI adoption curve is still steepening or if customer demand is softening. Second, the expansion of cloud revenue, particularly the infrastructure segment, will show if the company is successfully converting bookings into cash. Any deceleration in these numbers would be a red flag that the power bottleneck is already impacting execution. Conversely, continued strong growth would confirm Oracle's ability to navigate the build-out and solidify its position as a foundational infrastructure layer for the next paradigm. The path to 2027 is clear, but it runs through a bottleneck.