Good morning.
S&P futures flat, Nasdaq +0.2%. META +3% pre-market on the $10B Anthropic lease — validates their infra as monetizable third-party capacity. Memory names (MU, SK Hynix, Samsung) under pressure: US + South Korea regulatory probes into pricing signal PEAK CYCLE MARGINS. No major earnings prints overnight. Asia mixed — China’s Kimi K3 open-weight model (2.8T params, 60B active) narrowing the gap to US frontier labs, compressing model-layer margins but JEVONS PARADOX IN PLAY: cheaper inference drives more total compute demand.
Three themes framing today:
1. Model layer commoditization is a net positive for infra and software. Kimi K3 reinforces that model margins compress, but every dollar lost at the model layer flows downstream — infrastructure (compute, networking, power) and application-layer margins expand. DDOG, TWLO, CRWD all benefit as cheaper inference unlocks more workloads and agent deployment.
2. META-Anthropic deal confirms extreme AI compute scarcity. A hyperscaler selling capacity to a cash-rich lab — not a sign of oversupply. The world is short power and silicon. This underpins the TSM, AEHR bull case and keeps the data center capex supercycle intact despite rising funding costs (bond spreads widening on $1.8T off-balance-sheet AI debt).
3. Memory cycle peaking — be careful with the group. Regulatory probes are a warning shot. MU and Hynix face margin mean reversion as supply responds. That’s a headwind for high-beta memory names but neutral-to-positive for semi equipment (AEHR) as capacity adds keep going.
We’ll hit up DDOG, TWLO, and CRWD first (software tailwinds from cheaper inference), then get to TSM and AEHR in semis.
Verdict: Mixed signals but the AI narrative is re-rating the story. The stock sits at $332, up ~60% in a year, just off highs. You've got a HSBC upgrade to Buy ($366 target) vs. Jefferies and UBS neutral-to-cautious ($296-300). The bull case is real — 2.5B installed base, low capex leverage, product cadence. The bear case is real too — valuation at 40x, China AI approval called a "non-event," and iPhone 18 cannibalization risk. Earnings July 30 (Tim Cook's last call) is the next hinge point.
HSBC's upgrade is the loudest signal here. They see an "operational turning point" with the revamped Apple Intelligence platform — agentic Siri, cross-app context, on-device + private cloud models.
"Apple is at an operational turning point and is well positioned to leverage its 2.5 billion installed device base with its revamped Apple Intelligence platform." — Nicolas Cote-Colisson, HSBC
Key stat: Apple spends only 2.5% of 2026E sales on capex vs. 39% for hyperscalers. That's the beauty of the model — AI monetization without the infrastructure bill. Product pipeline supports the thesis: iPhone 18 Pro/Pro Max this fall, iPhone Air April 2027, a foldable, 20th-anniversary special, and smart glasses in '27.
BUT — Jefferies pushes back hard. China's AI approval? "Non-event" — government approval just confirms compliance. The real bottleneck is lack of app data for Apple Intelligence to work well. And Counterpoint data shows China iPhone growth was only ~7% in Q2, while Jefferies' own checks say China is down 4%.
June quarter looks benign. UBS sees iPhone rev of $53.3B (consensus $53.1B), total rev $107.8B (consensus $108.1B), EPS $1.84 (consensus $1.87). In-line, not a surprise driver. iPhone market share gained, Mac took share but at lower ASPs. Services growing ~13% FX-neutral but held back by App Store and Google search headwinds.
Bull case steelmanned: The installed base monetization through AI is a multi-year ramp. The product pipeline is the strongest in years — foldable, smart glasses, new form factors. HSBC raising 2027-28 revenue by 7-9% and EPS to $10.26 (7.5% above consensus) is a bet that this is not priced in. Evercore and Monness agree (Outperform, Buy).
Bear case steelmanned: At 40x P/E, you're paying for perfection. KeyBanc downgraded to Underweight ($250 target) citing declining iPhone spend and subsidy model shifts. Jefferies sees the iPhone 17 strength creating headwinds for the 18 — pulled-forward demand. The China AI narrative is a mirage without app access. And UBS is neutral at $296, implying ~11% downside from here.
If you believe the AI upgrade cycle is real and the product pipeline delivers, the HSBC $366 target is within reach (+10% upside). If you think this is a 40x stock with a mature phone market and no AI monetization catalyst for 12-18 months, you short it into the earnings print. The July 30 call — Cook's last — will be heavy on narrative, light on hard AI revenue. Watch the Services growth rate and China sell-through data more than the headline revenue beat.
UBS GOING TO $315. That’s the headline from Keirstead’s latest — Buy maintained, PT slashed UP 43% to $315 from $220. Stock at $259, up 90% last year. This is a “pay up for scarcity” call, not a valuation game.
Keirstead did the work — seven customers, three partners. Conclusion: fundamentals are strong, growth is accelerating into the 2Q26 print (early August), and there just aren’t many AI-exposed high-growth software names left to buy. The 122% six-month rip and 688x P/E? Yeah, that’s a tough setup. But UBS thinks the market will keep paying.
“Given strong fundamentals (confirmed by these checks), further growth acceleration in 2Q26 and a scarcity of AI-exposed high-growth software stocks, investors will remain willing to pay up for Datadog shares.”
THE WIDER TAPE: Benchmark went to $330, buying the product-led AI story. Bernstein downgraded to Market Perform — demand signals slowing into Q3. So you’ve got bull/bear in the same week. The Adaptive ML acquisition (RL ops) doesn’t move the needle near-term, but keeps the AI narrative alive. DASH conference pumped 1,000+ new features — noise or signal? The UBS checks say signal.
Verdict: This is a conversion schedule, not a pipe dream. Freedom Broker upgrades AEHR to Buy, PT to $110 from $90, after fiscal 2026 results and FY27 guidance of $130-150M. Record quarterly bookings of $60.7M pushed backlog above $100M – and they say every dollar is scheduled to ship this fiscal year. That’s the key distinction: no aspirational crap. Earnings beat was massive ($0.11 vs -$0.01 consensus), but the real signal is the backlog visibility. Analyst raised FY27 rev forecast to $133.2M and FY30 to $355.8M, implying serious compounding.
"The guidance represents a conversion schedule rather than a demand aspiration." — Freedom Broker
Stock already ripped 22% on the print. Still, at $110 PT, upside from current levels? Check the math. Bookings momentum + wafer test demand cycle = r/r worth a look for PMs who want semi cap exposure without the AI GPU noise.
BULLS ARE RUNNING THE TABLE – Mizuho, Stifel, Goldman, Rosenblatt all piling in with PT hikes. Mizuho takes target to $240 from $200 (Outperform), Stifel upgrades to Buy with $260, Goldman starts at Buy/$300, Rosenblatt stays at Buy/$230. The collective thesis: Twilio is critical AI-era comms infrastructure, messaging volumes healthy despite carrier price hikes, voice AI and agentic messaging adding volume growth. Q1 revenue +16% to $5.3B, and 18 analysts have revised EPS up. Shares already up 64% over the past year – Mizuho sees another ~16% upside from here.
"Twilio [is] critical communications infrastructure for the AI era, with voice AI and agentic messaging expanding the volume opportunity over the longer term." — Mizuho
Caveat: This is a crowded long now. Consensus conviction is high, but the rate of change on customer adds (half of Forbes 50 AI startups are paying) and organic guidance raise in Q3 are the near-term catalysts. If they print another above-consensus quarter, this thing can re-rate. If not, PT cluster becomes a ceiling.
Guggenheim leads the bull case here, hiking PT to $105 from $80 (21% upside) ahead of the August 6 print. The thesis is straightforward: cloud revenue re-acceleration (51% in Q2 vs 50% in Q1) and a guided raise — they expect 500bps bump in FY cloud revenue guidance to 38-40%. The catalyst is the AI-generated code wave and software supply chain security tailwind, with channel checks confirming demand. UBS and Benchmark have also chimed in at $110 and $100 respectively, so the street is consolidating around a ~$100-110 target zone.
"Conversations with chief information officers, senior engineering and security leaders, and reseller partners indicate JFrog continues to benefit from increased AI-generated code and heightened focus on software supply chain security."
At 77% gross margins and a sub-30% growth rate, the r/r is decent if they deliver the guided raise. The bull case works as long as cloud growth stays above 50% and the Anthropic Claude Code plugin proves more than a headline. Risk is execution on the margin expansion guide (21% op margin vs current 18% — that's a big jump). Not priced for error, but the narrative momentum is clearly shifting from "is the platform sticky?" to "how fast can they ride the AI development cycle?"
B.Riley flips back to Buy after six months on the sidelines. The catalyst? The $1B convertible note (1.625% due 2034) with a capped call that pushes effective conversion to $149 — a signal management sees room to run. Stock got slammed 44% since the January downgrade, now $55. Cash on hand blew past $2.7B post-Q2, and Riley models deployable cash hitting $3.4B by end of Q3. Next BlueBird batch (11/12/13) goes up early August, and they've got the first 45 satellites on track for early 2027.
"Riley models the recent offering as taking deployable cash to over $3.4 billion at the end of the third quarter of 2026, with the company expected to exit the period with $824 million of net debt."
Piper Sandler also came in overweight with a $100 target — no one is fighting the funding narrative anymore. Risk/reward shifted: the convertible overhang got a price floor, not a cap.
KGI RAISES PT TO NT$3,200 FROM NT$3,000, OUTPERFORM MAINTAINED. The call is simple: ignore the margin noise, buy the AI demand cycle. KGI’s Michael Liu acknowledges the "lack of upside surprises" in Q2 — but argues the real story is TSMC’s technology moat and a still-strengthening AI demand curve that drives a robust long-term earnings CAGR.
The new PT is based on 20x 2H27-1H28 EPS. That's a forward multiple that assumes the current earnings ramp is real and compounding. The firm is effectively saying the near-term gross margin compression (from higher capex and N2 ramp) is a temporary headwind, not a structural problem.
"While investors may be somewhat deflated by the lack of upside surprises in 2Q26 results, and may even harbor concerns about margin dilution (particularly from gross margin guidance), we suggest investors focus on the company’s sustained, outstanding technology leadership and how still-strengthening AI-related demand will drive a robust long-term earnings CAGR."
Key numbers from the print: Q2 rev of $40.2B (beat guide and consensus of $39.94B), gross margin 67.7% (above expectations), full-year revenue growth outlook raised to slightly above 40%. The bear case centers on margin dilution from the spending plans and lower Q3 margin guidance. But the PEG ratio of 0.7 is screaming that the earnings growth trajectory is underappreciated.
Positioning take: TSMC is the only game in town for leading-edge AI silicon. The stock trades at 20x forward EPS that is still rising. If you believe AI demand is a multi-year compounder, this is a structural buy. If you think the margin squeeze is the leading edge of a capex bubble, you wait. I lean with KGI — the rate of change in AI demand is still positive, and TSMC’s pricing power is intact.
Freedom Broker opens with Buy, $27 PT. Stock at $20.32 — already up 197% in the last year — so this is a “still room to run” call, not a discovery. Thesis hinges on the Security segment (+47% YoY) becoming a margin-mix engine, cross-selling WAF/bot management into the installed base, and the broader shift to high-margin software subs unlocking operating leverage and FCF. The Network Services tailwind (genAI, high-res streaming, API-heavy traffic) adds usage-based revenue per enterprise client.
The initiation lands against a backdrop where FSLY actually beat Q1 numbers ($0.13 EPS vs $0.08 est, $173M rev vs $170M) but got sold in after-hours — a reminder that good prints don't always save you when positioning is already stretched.
“Fastly is using its rapidly scaling Security segment, which grew 47% year-over-year, as a core potential margin-mix driver.”
Bet is that the mix shift from low-margin CDN to security/compute is real, the ARPU acceleration is durable, and the current multiple (still sub-$21) doesn't fully discount the FCF path. Risk: 197% run means the easy money is gone — any hiccup in growth rates and this thing gets repriced fast.
Stifel raises to $230 from $220, keeps Buy. Stock at $203.76, up 74% YTD, knocking on 52wk high of $217.50. The bull case is tightening: AI is creating a genuine demand inflection in cybersecurity, and CrowdStrike is the primary beneficiary. Multiple firms now targeting $230-240 (Freedom Broker $240, BTIG $237), with the collective thesis resting on Falcon Flex traction, FY27 net new ARR guidance upgrades, and C-suite urgency around AI-powered threats. Rosenblatt at $206 post-split is the outlier — ignore.
"AI is creating an inflection point in cybersecurity demand due to the risk that vulnerability discovery and exploitation can equip any hacker with the tools of a nation-state actor."
Management is also signaling M&A appetite (XM Cyber IP acquisition, no rev/customers — just patents). Bigger deals possible but high threshold to preserve single-agent architecture. The risk here isn't the story (it's working) but the valuation — 74% YTD leaves little room for a miss. That said, the rate of change in AI-driven conversations is accelerating, and PMs should watch for another FY27 guide raise on the next print.
Wells Fargo upgrades to Overweight from Equal Weight (PT trimmed to $210 from $220, but stock at $185 — so still 13% upside). Thesis is simple: domestic leasing bottoms H2 2026, improves through 2027-2028. The $3B debt raise kills near-term refi risk. AFFO yield vs. short-term debt cost has a ~200bp spread — buyback math is compelling, and Wells sees >$3B of headroom through 2029 at mid-6x leverage.
The take-private collapse at ~$230 (a 2-3x premium) and the broader towers selloff created the entry. PEG of 0.75 on 19.5x P/E is cheap if the leasing inflection actually materializes.
"Domestic leasing should bottom in the second half of 2026 and improve through 2027 into 2028."
Counterpoint: Q1 EPS missed by $0.04 (though revenue beat by $7M). Goldman started Neutral at $205 — not a ringing endorsement. The re-rating requires execution on the leasing narrative. Worth the risk/reward if you believe the trough is now.
OPERATING MANUAL IS CHANGING — and the market hasn't fully repriced the implications. Cantor stays Overweight at $28 (57% implied upside from $17.86), but this is less about the print and more about management's guidance philosophy shift. Q2 guide of $359-363M (23-24% y/y) looks soft vs Q1's 28% growth, but that's the point.
The explicit goal for 2026: guide closer to actual results. Q1 beat was only 2.7% vs historical 4-4.5%. Cantor estimates a similar Q2 beat prints $371M (26.5% growth) and suspects mgmt flows that ~$9M upside into full-year guidance, implying 23.7% expected growth vs current 23%.
"Management stated that for 2026 the goal is guiding closer to actual results than in prior years."
This is the narrative tension. Lower beat rates are a feature, not a bug — but it changes the r/r for PMs who got used to 4%+ beats as a margin of safety. Stifel already cut PT to $24 (maintains Buy) explicitly flagging this will be tested next earnings. Goldman initiated at $26 Buy.
Bull case: stable 23-25% grower with expanding product surface (new Social Marketing integration is real). Bear case: guide-down in beat rate creates a dead zone for momentum players until the new cadence is established. KVYO needs to print the Q2 beat and show the flow-through before the stock re-rates.