TMTB Morning Wrap
TMTB will be on vacation from 6/30 - 7/11. Tomorrow will be our last post before we leave
QQQs +40bps looking to set more ATHs as AI semis strong to start. BTC flat. China flat. Yields flat.
Lots to get to this morning, so let’s dive straight in…
MU: Solid beat and raise across the board with some things to nitpick…bulls remain in control
Despite high expectations and a >100% run off the bottom, this print should be enough to keep bulls happy, although some nitpicks investors debating this morning: 1) they admitted tariff pull-ins could be a factor in strong #s, 2) wouldn’t commit to say their “sold out” for HBM in CY26; 3) NAND responsible for significant portion of the beat. However, we think magnitude of the beat and raise enough to outweigh those issues although stock could digest some today.
The #s:
Q4 EPS $2.35-$2.65 vs bogeys of 2.20-2.30 - highest I had heard was 2.50
Q4 Revs $10.4B - $11B vs street at $9.9B, above bogeys
Q4 GM 41-43% vs street at 39% inline/better than bogeys
Q3 Revs $9.3 vs street at $8.85B
Q3 EPS $1.91 vs street at $1.60
Q4 GM 39% vs street at 36.7%
Key Takeaways:
On tariffs, mgmt acknowledged tariff pull-ins “may” be a factor and any new tariffs not factored into guidance.
HBM revenue rose “nearly 50 % q/q,” reaching a >$6 B run-rate; management targets 20 - 25 % share by 2H-25 and sees 2026 HBM bit growth outpacing DRAM industry demand. Jefferies said on the analyst callback, mgmt basically admitted that they don’t have full visibility currently on ‘26, as customers are going thru rapid transitions and are still in the process of finalizing their 2026E plans
Margin expansion driven by mix and pricing. Management attributed the 110 bp sequential GM lift — and another 300 bp implied in guidance — to “better-than-expected pricing,” richer DRAM mix and cost/scale gains from the HBM ramp. The one nitpick is incremental margins lower than expected in guide with DRAM to out-grow NAND
Inventory rapidly normalizing. Days-in-inventory fell 19 days q/q to 139; management expects company-wide DOIs “near target” exiting FY-25, underpinning firmer pricing
Capex tilted to AI memory. FY-25 capex unchanged, overwhelmingly for HBM capacity and advanced packaging; Boise “fab 2” and Virginia expansion highlighted as part of a longer-term $200 B US investment plan.
Bull vs. Bear Debate:
Bulls argue that cyclical and secular are aligned and that old-time semi-investors don’t appreciate the magnitude of the current cycle where higher HBM demand is helping drive higher commodity DRAM pricing. Bulls will say MU is the pure-play on AI memory scarcity and #2 beta-levered way to play “AI vibes” + Blackwell launch + AI demand. The the HBM ramp is ahead of plan, the company is now the #2 data-center SSD vendor, and tight inventories plus disciplined wafer-start growth and mix shift to HBMs should sustain a multi-year gross-margin climb above 40 %.
Bears counter that the stock already discounts much of the AI upside with stock trading near peak P/B ; HBM share gains could slow as Samsung re-enters, and spot pricing can “turn quickly if inventories creep up.” as well as tariff pull ins driving some of the demand. Bears will say NAND remains structurally softer, and any abrupt tariff escalation or cyclical PC/mobile weakness could squeeze margins back below the newly set bar.
We think the alignment of both cyclical and secular tailwinds is the most potent point and the print and commentary confirmed that. Expectations were high and a few things to nitpick on the print, but we think bulls remain in control: Bulls close to $14-15 in ‘26 and will put 13x = $180+
TECH RESEARCH/NEWS
MRVL: Jefferies Asia Sees MSFT Retention, AAPL Win Potential in ASIC Pipeline
Jefferies Asia now believes MRVL is likely to retain Microsoft's 2nm Mia 2.0 ASIC, reversing earlier speculation that Broadcom (AVGO) could win the business. Additionally, Apple, which developed its first ASIC with Broadcom, is now evaluating Marvell for its second-generation chip due to cost concerns. While the AAPL selection isn't final, Jefferies notes Marvell is in serious contention if it can prove performance parity with AVGO. Jefferies think xAI dual sourcing AVGO/MRVL now while OpenAI AVGO delayed.
Continued good news flow here from Asia helping investors get more comfortable with what MRVL mgmt has been saying…MRVL getting its groove back means the ‘23/’24 horsemen of NVDA, AMD, MU, MRVL, AVGO all back in the saddle
U: BofA Reinstates at Underperform, $15 PT; Questions Monetization and Premium Valuation vs. Ad Tech Peers
BofA reinstated coverage of Unity Software (U) with an Underperform rating and a $15 price target, arguing the company’s flagship game engine offers limited monetization potential despite widespread usage. The firm notes that while ~70% of mobile games run on Unity, only ~10% of developers pay for the engine, and its seat-based subscription model doesn't drive strong financial returns. Unity's ad business (Grow) has lagged peers since 2022, and while game engine data could eventually boost targeting and ad revenue, BofA remains unconvinced without clear traction. The firm sees potential upside if Create or Grow segments inflect meaningfully, but current valuation at 17x CY26E EBITDA appears rich relative to comparable online ad peers trading at 9x–20x. Without material revisions to forecasts, BofA expects Unity’s multiple to compress closer to smaller-scale ad peers.
U: In Piper’s APP note this morning, they note that that Advertiser checks still show improving sentiment around Unity Ads.
TTD: Wells Fargo Cuts PT to $68, Downgrades to Equal Weight on Amazon DSP Gains
Wells Fargo downgraded The Trade Desk to Equal Weight from Overweight and lowered its price target to $68 from $74, citing rising competitive pressure from Amazon DSP following new Roku and Disney deals. Advertiser checks show Amazon gaining wallet share in non-Amazon CTV inventory, with 30% seeing it take "significant share." While 2025 estimates remain achievable, Wells trimmed FY26/FY27 revenue and EBITDA by 2%/5%. Kokai is showing some performance upside, but user interface concerns persist, with advertiser preference now evenly split with Solimar. Valuation at ~20x EV/’27 EBITDA limits further upside.
RDDT: Reddit vows to stay human to emerge a winner from artificial intelligence
Some good quotes from Cannes from FT around “massive migration” to platform…
Keep reading with a 7-day free trial
Subscribe to TMT Breakout to keep reading this post and get 7 days of free access to the full post archives.