Nvidia and Salesforce Earnings Shatter Bear Narratives as AI Capex Cycle Accelerates Under Geopolitical Strain

INTRODUCTION

The technology landscape on August 28, 2026 is defined by a striking juxtaposition: accelerating enterprise AI adoption validated by blockbuster earnings from Nvidia and Salesforce, set against a backdrop of escalating geopolitical tension as the United States pursues expanded sanctions against Iran and seeks G20 coordination on enforcement. Salesforce surged roughly 20 percent in a single session—on track for its second-best trading day in history—after reporting a strong beat on second-quarter earnings and announcing an expanded AI partnership with Anthropic. Nvidia's results simultaneously dismantled the bear thesis that AI infrastructure spending was peaking, confirming that hyperscaler and enterprise demand for accelerated compute remains structurally robust. Meanwhile, Federal Reserve Chairman Kevin Warsh's forthcoming remarks at Jackson Hole are expected to avoid explicit discussion of rate cuts or the bond market, signaling a Fed still calibrating its stance amid persistent fiscal uncertainty and energy-price volatility linked to Strait of Hormuz disruptions. These threads—AI monetization, semiconductor demand, monetary policy, and energy security—converge to shape the investment and strategic calculus for every major technology stakeholder.

FUTURE PROJECTIONS

BEST CASE:

Salesforce's Anthropic partnership becomes a template for enterprise AI co-development, driving durable 25-plus percent growth in the company's AI-attached revenue streams while validating the broader SaaS sector's ability to monetize large language model integrations. Nvidia's data-center GPU shipments continue to exceed consensus through fiscal 2027 as sovereign AI programs, hyperscaler buildouts, and enterprise inference workloads layer new demand on top of training demand. The G20 achieves partial consensus on Iran sanctions, preventing a prolonged closure of the Strait of Hormuz and stabilizing energy input costs for data-center operators. The Fed holds rates steady, providing a favorable discount-rate environment for growth equities. Under this scenario, the Nasdaq could retest all-time highs within two quarters.

BASE CASE:

AI monetization proves uneven across the software stack. Salesforce sustains elevated growth but faces margin pressure from Anthropic partnership economics, and competitors like Microsoft Copilot and Google Workspace AI narrow the differentiation gap. Nvidia maintains strong revenue but faces emerging competition from AMD MI400 and custom ASIC alternatives from Amazon (Trainium3) and Google (TPU v6). Sanctions enforcement fragments along geopolitical lines, with China and several Global South nations declining to comply, creating a bifurcated energy and semiconductor supply environment. Warsh signals data dependence, keeping markets range-bound. Enterprise IT budgets grow mid-single digits, with AI projects cannibalizing legacy spend rather than expanding total addressable market.

WORST CASE:

Strait of Hormuz disruptions escalate, driving sustained energy price spikes that raise data-center operating costs 15-20 percent and compress margins across the cloud and AI infrastructure stack. The Fed is forced into a hawkish posture to combat energy-driven inflation, raising discount rates and punishing high-multiple tech equities. Enterprise buyers defer AI deployments amid macro uncertainty, exposing the gap between AI capex commitments and near-term revenue realization. Nvidia faces inventory correction risk if hyperscaler orders moderate, echoing the 2022 crypto-driven GPU oversupply cycle.

HISTORICAL CONTEXT

The current AI capex cycle echoes the 2005-2008 cloud infrastructure buildout, when Amazon Web Services launched amid skepticism about whether enterprises would outsource compute. That cycle proved durable but nonlinear, with the 2008 financial crisis temporarily compressing demand before secular adoption resumed. Nvidia's trajectory mirrors its own 2016-2018 deep-learning inflection, when data-center revenue grew from a rounding error to a primary business driver—only to suffer a sharp correction in 2018-2019 when crypto demand evaporated. The key structural difference today is that AI workloads are anchored to enterprise productivity use cases with measurable ROI, rather than speculative mining. Salesforce's Anthropic partnership extends a lineage of platform-model integrations dating to Microsoft's OpenAI investment in 2023, which reshaped competitive dynamics across the SaaS landscape and forced every major software vendor to secure foundation-model access.

PRIMARY STAKEHOLDERS

Nvidia remains the indispensable supplier of accelerated compute, but its moat depends on sustaining CUDA ecosystem lock-in against increasingly capable alternatives. Salesforce is positioning itself as the enterprise AI distribution layer, leveraging its CRM installed base of over 150,000 customers to channel Anthropic's Claude models into workflow automation. Anthropic gains a scaled go-to-market partner, reducing its dependence on direct API revenue. Hyperscalers—Amazon, Microsoft, Google—face a tension between partnering with model labs and building proprietary alternatives. The Fed under Warsh must balance financial stability against inflation risk amplified by geopolitical energy shocks. Enterprise CIOs are the swing factor: their willingness to commit multi-year AI budgets will determine whether the capex cycle extends or contracts.

ECONOMIC IMPLICATIONS

Nvidia's earnings confirm that data-center GPU demand, particularly for H200 and Blackwell-generation products, remains supply-constrained rather than demand-constrained—a critical distinction for semiconductor supply chain planning. Salesforce's 20 percent single-day move reprices the entire enterprise SaaS sector's AI optionality, potentially compressing the valuation discount that software stocks have carried relative to semiconductor plays since 2023. If Salesforce's Anthropic deal economics prove accretive, expect a wave of similar partnerships across ServiceNow, Workday, and SAP. Energy cost volatility from Iran-related disruptions introduces a new variable into data-center site selection and total cost of ownership models, potentially accelerating nuclear and renewable power purchase agreements among hyperscalers. The net effect is a technology economy increasingly bifurcated between AI winners capturing disproportionate value and legacy vendors facing accelerated displacement.