☀️ Morning Briefing
Highlighted Companies
KEY COMPANIESServiceNow beat every top-line Q1 metric and raised full-year Now Assist ARR guidance to $1.5B — 50% above prior forecasts — yet shares cratered 18%. The market isn’t punishing the results; it’s discounting the category. Investors appear to be pricing in structural risk that agentic AI erodes the platform-layer moats that ServiceNow’s premium multiple depends on.
Read More →Google’s Cloud Next messaging crystallized around control-plane dominance — positioning Gemini as the orchestration layer across IDEs, notebooks, and agentic terminals, with Salesforce, SAP, and ServiceNow all deepening integrations. The framing puts Google directly in competition with AWS Bedrock, Microsoft Copilot, and Salesforce Agentforce — all converging on the same architectural bet that the value in enterprise AI accrues to whoever controls the workflow layer.
Read More →Salesforce shed 9% on the ServiceNow-triggered selloff — not on its own fundamentals. Agentforce closed FY26 at $800M ARR (up 169% YoY), with 29,000 deals signed and three parallel pricing models in market. The dislocation between execution and stock performance reflects macro-level fear about AI SaaS disruption, not Salesforce-specific risk — which may create a near-term entry opportunity.
Read More →The ServiceNow reaction detonated across enterprise SaaS: Workday, now down 45% year-to-date, absorbed another 9% drop; Adobe and HubSpot followed. The common narrative is that AI agents will erode seats-and-licenses economics that underpin SaaS multiples. The divergence between AI infrastructure stocks (Intel +24%, NVIDIA +4%) and application-layer SaaS is becoming the defining trade of 2026.
Read More →Five of the Magnificent Seven report this week against a backdrop of S&P and Nasdaq at all-time highs. The market’s resilience will be tested by whether hyperscaler capex commentary (Amazon, Alphabet, Microsoft) confirms or cools the AI infrastructure thesis. Powell’s FOMC meeting adds a macro overlay — any signal on rate trajectory lands while valuations sit at elevated multiples.
Read More →AI & Semiconductors
AI & CHIPSIntel reported Q1 revenue of $13.6B against estimates of $12.4B, with Data Center & AI at $5.1B (vs. $4.4B expected) — AI business up 40% YoY, sixth consecutive quarter beating guidance. Q2 guidance of $13.8B–$14.8B is $800M above consensus. This is Intel’s credibility inflection: the turnaround narrative is no longer aspirational, it’s in the numbers.
Read More →Broadcom locked in expanded TPU and custom accelerator agreements with both Google and Anthropic, giving Anthropic access to ~3.5GW of compute capacity. Separately, Broadcom is co-developing 10GW of custom AI accelerators with OpenAI targeting H2 2026 deployment. Broadcom is quietly becoming the custom-silicon kingmaker as hyperscalers diversify away from NVIDIA — a structural overhang for NVDA at the margin.
Read More →Meta formalized a 1-gigawatt custom chip commitment with Broadcom — the largest single custom-silicon deal yet disclosed — as CEO Hock Tan agreed to leave Broadcom’s board, resolving a potential conflict-of-interest concern. The scale of the deal underscores that hyperscalers are not supplementing NVIDIA with custom silicon; they are structurally replacing meaningful portions of their inference stack with it.
Read More →NVIDIA’s Rubin platform — launched at CES, now moving into partner production for H2 2026 availability — features 336B transistors, 288GB HBM4 at 22TB/sec bandwidth, and claims a 10x reduction in inference token costs vs. Blackwell. AWS, Google Cloud, Microsoft Azure, and OCI are first-wave deployers. NVIDIA’s $1T total Blackwell+Rubin revenue forecast for 2026–2027 sets the stakes for the custom-chip challengers.
Read More →AMD, Broadcom, and Intel have all secured deployment commitments from major AI labs and hyperscalers — a structural shift from NVIDIA as sole supplier to a diversified silicon stack. NVIDIA retains dominance in training and frontier model development, but the inference layer — where the volume and economics are — is increasingly running on custom ASICs. The real threat to NVIDIA isn’t AMD; it’s Google TPUs, Meta MTIA, and Amazon Trainium at scale.
Read More →Markets & Tech Stocks
S&P 500 · NASDAQ · MARKETSEquity markets closed Friday at record levels: S&P 500 +0.80%, Nasdaq +1.95%, driven by Intel’s blowout and renewed optimism on US-Iran diplomacy. The S&P is up more than 9% in April alone — an extraordinary single-month move — with 80% of the 139 reporting S&P 500 companies beating Q1 estimates. Q1 earnings growth is tracking at +12% YoY, but strip out tech and that figure collapses to ~3%.
Read More →The hardware/software divergence widened dramatically: Intel posted its best single session since 1987 (+24%), AMD and ARM each gained 13%+, while enterprise SaaS bled out on AI displacement fears. The market is telling a coherent story — AI infrastructure capex is a multi-year spend cycle, but platform-layer SaaS economics are genuinely under threat. The spread between these two cohorts is becoming the defining factor allocation trade of 2026.
Read More →Amazon, Alphabet, Microsoft, Meta, and Apple all report this week — with the Fed meeting running concurrently. Hyperscaler capex commentary will either validate or complicate the AI infrastructure thesis that’s driving semiconductor multiples. Powell’s meeting, potentially his last, carries rate-path implications that will land against a market pricing in cuts by mid-year. Any miss from a Mag 7 name at current valuations could trigger outsized moves.
Read More →The US-Iran ceasefire extension removed a key macro risk premium that had been embedded in equity valuations since the conflict escalated. The relief rally compounded momentum from strong tech earnings, pushing indices to fresh highs. ServiceNow had explicitly cited the US-Iran war as a headwind in its guidance — meaning any further diplomatic progress could serve as a direct sector-level tailwind for enterprise software demand.
Read More →FactSet’s mid-season update puts Q1 EPS growth at +12% with an 80% beat rate — healthy headline numbers masking a significant concentration story. Ex-technology, S&P 500 earnings growth drops to ~3%, the weakest in two years. The market is increasingly a two-speed economy: AI infrastructure and enabling hardware versus everything else. That bifurcation has direct implications for how enterprise buyers prioritize budget in the back half of the year.
Read More →Supply Chain & Commodities
CHIPS · MATERIALS · FREIGHTSemiAnalysis estimates memory will account for ~30% of hyperscaler capex in CY2026, up from ~8% in CY2023 and CY2024. The driver is HBM — NVIDIA reportedly receives preferential supply terms well below standard market rates, creating a two-tier pricing environment that disadvantages smaller AI labs and non-hyperscale cloud operators. Memory is no longer a commodity input; it is a strategic constraint on who can compete in frontier AI.
Read More →The big-five hyperscalers (Amazon, Alphabet, Microsoft, Meta, Oracle) are collectively projected to spend $600B–$690B on capex in 2026, up ~36% from 2025 — with ~75% ($450B+) tied directly to AI infrastructure. Capital intensity has reached 45–57% of revenue, historically unthinkable, with hyperscalers increasingly using debt markets to bridge the gap between free cash flow and AI ambition. The spending signals are a direct accelerant for NVIDIA, Broadcom, SK Hynix, and Micron.
Read More →DRAM pricing accelerated sharply in Q1 — following 30% increases in Q3 2025 and 40–50% in Q4 — as HBM production consumes 3–4x the wafer capacity of standard DDR5 per gigabyte, structurally tightening supply. Micron, SK Hynix, and Samsung are managing demand through allocation-only frameworks favoring hyperscalers and large OEMs, effectively rationing supply and locking in preferred buyers at the expense of everyone else.
Read More →SK Hynix, NVIDIA’s primary HBM supplier, has communicated to investors that its advanced packaging lines are fully allocated through end of 2026 — meaning any incremental AI compute demand gets queued, not fulfilled. Micron’s new Idaho fab won’t reach operational status until 2027. Combined capex commitments (Micron $13.5B, SK Hynix $20.5B, Samsung $20B) represent a massive build cycle, but supply relief is structurally 12–18 months away.
Read More →Despite explosive demand, semiconductor supply chain analysts flag a paradox: foundry and advanced packaging capacity expansions remain conservative, with no major player aggressively scaling fab output. The risk is that the current allocation crunch extends further than anticipated, creating a structural mismatch between hyperscaler capex ambitions and the physical supply chain’s ability to deliver. This dynamic is sustaining elevated pricing and keeping margins high for HBM producers through the planning horizon.
Read More →