LT Weekly: Jul 17, 2026

Q2 was a similar narrative for Netflix as in Q1, given that the Co delivered a give-or-take in-line qtr with lowered expectations, while the Q3 guidance fell below Street projections and 2026 was largely maintained (though in this case with a narrowed revenue range). Aside from the decelerating top-line growth implied in the Q3 targets, investors were also not thrilled that Co is reducing engagement related disclosures going forward.

The latter point adds fuel to the fire on general engagement concerns, which have been an overhang for NFLX shares. Mgmt continues to reiterate that viewing hours are only one component of measuring engagement, given that all hours are not created equally. With that said, mgmt highlighted that view hours grew +2% y/y in H1:26 (up +1.5bn hrs to 97bn+) which is a slight accel vs +1.5% in 2025, despite competition from the Winter Olympics and World Cup. They also dismissed press reports that viewership of Season 2s was dramatically falling off from Season 1s, and in fact indicated that the Season 2 falloff has “actually slightly improved this year relative to last year.”

Aside from that, live remains an increasingly important component of Netflix’s content portfolio, as does new formats like video podcasting and cloud gaming. The TF1 partnership, while early, is showing positive signs and it wouldn’t be surprising to see similar deals on this front looking ahead. Mgmt reiterated that content amortization remains front-end loaded and overall content spend will rise +~10% y/y this year.  Lastly, the doubling of the ads business also remains on track.

See more color below on the takeaways we found most impactful, including some comments on AI usage in production.

-> Netflix share fell -7.3% in reaction to earnings (following falling almost -10% in response to Q1 results); The stock is now down -26.5% YTD

It Was Hard to Get Too Excited About Q2 Results ….

• Q2 revs were in-line and grew by +13.4% y/y (+12% y/y FXN): Growth was driven primarily by membership growth, pricing and incr’d ad rev, with DD rev growth across all regions
  • UCAN disappointed: Grew +10.2% y/y and MISSED cons by -1.5% as Q2 only included a partial qtr impact from recent price changes (which “has gone well and as expected)
  • LATAM outperformed: Grew +20.9% y/y and BEAT cons by +4.6%
  • EMEA grew +13.9% y/y and was in-line with cons
  • APAC grew +15.7% y/y and MISSED cons by -1.3%

• Op margin of 33.4% BEAT cons 32.8% and was ahead of the 32.6% guide given timing of expenses
  • BUT op margin was down ~70bps y/y, reflecting higher content amortization growth in H1, which mgmt expects to moderate in H2

• Adj EPS BEAT cons by +1.3% and grew +11.1% y/y: Driven by op income beat and lower share count

• FCF MISSED cons by -39.0% ($1.53bn vs cons $2.51bn) and decr’d -32.5% y/y; Driven by higher cash tax payments due in part to the Warner Bros. termination fee
• Repurchases accelerated in Q2…$4.7bn buyback was the Co’s largest qtr of repurchases and the Co has $27.1bn left on the authorization

…AND Q3 Guidance Disappointed While The FY Guidance Was Narrowed

• Q3 guidance MISSED cons across rev / op income / adj EPS…and implies another qtr of decelerating FXN growth despite healthy membership / pricing / ads drivers
  • Rev guide of $12.86bn MISSED cons by -1.1%…implies +17% y/y growth (+11% FXN) vs +13.4% in Q2
    • Mgmt attributed the q/q decel to q/q choppiness b/c 2025 was more H2 weighted and emphasized “we manage to the full year”
  • Op margin guide of 33.2% MISSED cons 33.5% BUT was up ~500bps y/y (vs 28.2% in Q3:25)
  • EPS guide of $0.82 MISSED cons of $0.84

• 2026 guidance was NARROWED around the prior mid-pt
  • The rev range NARROWED to $51.0-51.4bn (from $50.7-51.7bn) w/ the midpt held at ~$51.2bn (-0.4% vs cons)
    • Implies +13-14% y/y growth (~12% FXN)
    • Growth will be driven by membership growth, pricing and ad revs doubling to ~$3bn
  • Op margin guidance of 31.5% was MAINTAINED
    • Implies +20%+ annual op income growth vs 2025
  • FCF guidance of ~$12.5bn was MAINTAINED
  • Annual cash content spend-to-amortization remains at ~1.1x

Mgmt Defends Its Engagement KPI BUT Also Reduces Going Forward Disclosure Cadence, Keeping The Debate Front & Center

• View hours grew +2% y/y in H1’26 (+1.5bn hrs, to 97bn+), a slight accel vs +1.5% in 2025…despite competition from the Winter Olympics and World Cup

• Mgmt reiterated again that regarding engagement it is not just about viewership hours as “all hours are not created equal”
  • It is about quality, variety and quantity… different content types drive acquisition / retention / monetization
  • Non-English content again drove >1/3 of viewing…standouts from Korea, Japan, Spain, and India

• Mgmt also pushed back on analyst concern that 2nd Season viewing is weakening vs 1^st Season
  • The Season 2 falloff has “actually slightly improved this year relative to last year”
  • Mgmt said there will be “no changes in release strategies”

• At the same time mgmt highlighted its strong engagement stats, it announces that it is REDUCING its engagement disclosure cadence
  • After this H1:26 What We Watched report, it will shift to annual (in Q1) publication starting in 2027
  • Mgmt explained the change as keeping “the focus on our primary financial metrics – revenue and operating profit” …Co will keep weekly Top 10s and title-level hours

Content Spend Wil Grow +~10% Y/Y In 2026 BUT Content Amortization Remains Front End Loaded

• Content expense is expected to grow ~10% in 2026, above the 5-yr avg of +8% but below the decade avg of +14%
  • Mgmt emphasized that spend grows slower than rev and the Co remains disciplined
  • Content amortization growth is H1 weighted and should decelerate to mid-to-high SD-digit growth in H2

• Core TV / film remains the vast majority of programming spend and the Q2 slate breadth was strong
• Q3 slate includes 72 HOURS, The Last House, The Whisper Man, Call My Agent!, The Hawk, Outer Banks final season, Little House on the Prairie, Monster: The Lizzie Borden Story, The Gentlemen S2, The East Palace, The Doll and Lovesick

Live Events Are Becoming Increasingly More Important

• Live is a disproportionate acquisition / engagement driver…expected to be ~5% of content spend, but only ~1% of view hours in 2026 as mentioned above
  • BUT 6 of the top 10 new member sign-up days over the past 5 yrs came from live events (Co has only run live since 2023)
  • In comparison, animation series kids family TV is also ~ 5% of content spend, but will drive 8% of view hours

• Live strategy remains focused on “big breakthrough events” vs regular-season packages

• World Baseball Classic (WBC) in Japan became NFLX’s most-watched program ever in Japan and the biggest baseball streaming event ever
  • WBC-driven members showed churn in-line with mgmt’s expectations despite slightly higher churn characteristics from event-driven acq

• Key upcoming live events –
  • NFL: A week-one matchup in Q3, a Thanksgiving Eve game and NFL Christmas Gameday in Q4, and a final-week contest in Q1’27
    • Mgmt said learnings from the last couple of yrs will make them “even more disciplined” in how they value the NFL and live generally
  • MLB Home Run Derby and MLB Field of Dreams, Tyson Fury vs Anthony Joshua, Women’s World Cup, WWE

Expect The Co To Continue To Expand The Variety Of Its Content

• Early results of the TF1 partnership in France (launched last month) is “very promising”…TF1 view hours are growing weekly and Secret Story already hit the France Top 10
  • The Co is open to similar types of deals but there is “nothing new to announce” now

• Video podcasts show incremental engagement (viewership over-indexes in the daytime + on mobile)
  • The Co added Jay Shetty’s On Purpose, Kate & Oliver Hudson, plus publisher lifestyle content via Condé Nast, Hearst, and People (starting Aug) and iHeartMedia

• Creator and publisher relationships are becoming a bigger part of the engagement mix
  • Successes include Danny Go!, Ms. Rachel, Mark Rober, Salish & Jordan Matter; annc’d collaborations include Stokes Twins, Alan Chikin Chow, Nick DiGiovanni and Mythical
  • Publisher partnerships with Condé Nast, Hearst and People will bring lifestyle content to members in the US and several other countries beginning in August

• Cloud gaming is gaining traction…FIFA World Cup & Unhinged were the two most successful cloud-game debuts (June)…cloud MAUs are up +11x since Oct, w/ adoption ahead of the curve that they had for mobile games, and w/ even higher retention
  • The Co is also seeing positive signals w/ kids games: Netflix Playground is posting +3x daily players since April, driving kids mobile-game engagement +600% y/y
  • Mgmt think they are still scratching the surface: Cites a ~$150bn market opportunity (ex-China/Russia)
  • Investment remains small relative to total content spend and will be calibrated to demonstrated member performance / business returns

Advertising Revs Remain On Track To Double To ~$3bn In 2026…ARM Gap Is A Key Monetization Opportunity

• Mgmt reiterated ad rev is on track to roughly double y/y to ~$3bn in 2026
  • US upfront negotiations are “in advanced stages”…commitments are expected to close “in the next few weeks”
  • There is strong advertiser interest in the Live lineup, i.e., see above
    • Plus the breadth of entertainment titles

• Ad-tier ARM remains below Standard w/out Ads ARM, BUT the gap is narrowing and mgmt called it “near-term underrealized” rev growth
  • The main levers are expanded demand sources, faster execution on Netflix Ads Suite, more ad products, more measurement, easier buying, higher fill rates and higher ads ARM
  • Those unit monetization improvements are the “bulk of the opportunity” for the next few yrs

• Other key updates on the ads business
  • Programmatic continues to expand beyond non-live inventory
    • NFLX is extending programmatic access to Pause Ads and live inventory this summer, which should reduce manual buying friction and open inventory to a broader advertiser base
  • The Co expanded AI-powered tools across the full ad lifecycle (planning, creative, campaign mgmt, optimization, reporting

A Few Other Important Updates…

• First-half price changes (US, Mexico, Spain) “have gone well”…the impact is “consistent with prior price changes and our expectations,” w/ no change in receptivity this cycle
  • Mgmt argues that US subs pay “the least per hour of viewing” vs comparable SVOD (“in some cases…2x per hour” for competitors)…the $8.99/mo ad plan framed as an “amazing entry point”
  • Mgmt also said a free offering could make sense in some markets, BUT highlighted potential paid-tier cannibalization, offering differentiation, and scaled ads economics as a gating factor

• GenAI workflows have been used in roughly 300 titles in 2026, with the largest concentration in post-production…InterPositive is now working alongside iLine and the animation lab
  • AI has enabled complex shots/sequences productions that they “would have left out”
    • Example: the American Experiment doc had 17 min of AI-enhanced footage produced “twice as fast and at half the cost”
  • Cost savings are “likely reinvested into more content” rather than cutting the cash content budget
  • “It takes great artists to make something great…AI is not changing that”

• No interesting comments on M&A…the bar for large M&A remains “very high” and mgmt emphasized no change in strategic profile / capital allocation

In a deal that has been a long-time in the making, Uber finally came to an agreement to buy the remaining shares of Delivery Hero that it does not own for €41.50/shr in cash (a ~34% premium to the 3-mo volume-weighted avg shr price). That price implies an enterprise value of $14.8bn. As a reminder, Uber was already Delivery Hero’s largest shareholder and had an earlier takeover approach at €33/shr which was rejected back in May, but this sweetened bid did the trick. The combination of the two companies would make it the largest food-delivery group outside of China’s Meituan, and this deal marks the latest consolidation of the food-delivery market, following DoorDash’s £2.9bn takeover of UK-based Deliveroo and Prosus’s €4.1bn acquisition of Just Eat Takeaway last year.

Concurrently, Delivery Hero is selling operations in 14 overlapping markets to SSW Partners for $1.6bn. If the deal closes (expected in H2:27), Uber would essentially double its geographic footprint to 99 mkts with combined total gross bookings of $236bn. The deal also nearly doubles the number of markets where Uber offers both mobility and delivery from 34 to 58, unlocking over 50mn new eligible cross-platform users. Annualized synergies are expected to exceed $1.2bn within 18 months, though those estimates are characterized as “very conservative” with high confidence they can be exceeded.

See below for more color on the deal and opportunities in cross-platform, advertising, tech integration, the regulatory path, the competitive landscape, and more (link/link/link/link).

-> Uber closed the day of the deal announcement up +0.8%, while Delivery Hero was up +0.9%; YTD, Uber is still down -11.3%, while Delivery Hero is up +68.2%

• The deal is not a huge surprise… Uber was already Delivery Hero’s largest shareholder and had an earlier takeover approach rejected in May, with press speculation that they were still trying to get a transaction done…

• Deal basics:
  • Price/shr: €41.50/shr of cash
    • A ~34% premium to Delivery Hero’s ⁠3-mo volume-weighted avg shr price
    • That also compares to Uber’s earlier proposal of €33/shr back in May
  • Valn: Implied EV is $14.8bn (or $13.7bn adj for Uber’s prior stake purchases)
    • Multiple: Implies ~8x EV / 2027E adj. EBITDA (incl. existing ownership and $1.2bn+ of run-rate synergies)
  • Financing: Uber will fund the deal through existing cash on its balance sheet and new debt financing
    • The Co is committed to maintaining its investment grade rating and to its capital allocation framework
    • Gross leverage is expected to remain below 2x
  • Timing: Expected to close in H2:27, pending regulatory approvals

• Alongside this transaction…Delivery Hero struck a separate $1.6bn deal to sell its biz in 14 countries to SSW Partners, a NY -based investment group
  • The sale covers mkts where Uber Eats and Delivery Hero already overlap, and is subject to the larger takeover deal’s completion

• Synergy realization –
  • Expect annualized synergies of $1.2bn+ w/in 18 months of closing, though mgmt. is “highly confident” they can do more than that
  • The transaction is expected to be non-GAAP EPS accretive upon close, and HSD %age accretion by yr 3

• What are the biggest synergy drivers? Migrating Delivery Hero’s brands onto Uber’s global tech platform is the “first and biggest” synergy driver
  • Despite Delivery Hero operating w/ a net take rate that’s higher than Uber’s, its margin output is “significantly” lower than Uber’s…and one of the biggest deltas is the cost of tech on a % on gross bookings basis at Delivery Hero relative to Uber’s
  • Integration complexity is reduced b/c both Uber Eats and Deliver Hero (ex-Baemin in Korea) already run on common back-end architectures, making this a platform migration rather than a multi-yr re-platforming effort
  • Addtl cost savings are expected “pretty quickly” from shared svs (cost of payment, support, insurance, etc.) b/c of the migration

• On the regulatory front, mgmt sees a “clear path to closing,” as the transaction has been structured to facilitate the regulatory process while preserving the strategic value of the combination
  • The deal is fundamentally about expanding Uber into “highly complementary” mkts rather than combining two delivery bizs everywhere (i.e., no delivery-on-delivery overlap given the SSW carve-out of 14 overlap mkts)
  • While the German takeover process is complex, it involves “well-defined steps” that have been “successfully navigated by many international acquirers before”
  • “We’re quite confident in both the strategic merits of the transaction and then, of course, the path to completing it”

• Total mkts – the takeover will expand Uber’s mobility and delivery platform from 79 -> 99 mkts, creating a group with combined gross bookings of $236bn
  • The deal would expand Uber Eats across Europe, the Middle East, Asia and Latin America

• The Co’s expanded cross-platform position unlocks “meaningful” topline synergies, with the deal growing the # of mkts where Uber offers both mobility and delivery from 34 -> 58 mkts
  • “Cross-platform work is one of the highest return growth levers that we run on a global basis. There’s very little investment. We’ve already acquired these customers. All we’re doing is cross-selling each other”
  • Unlocks 50mn+ new eligible cross-platform users: ~35mn from Delivery Hero’s delivery user base + ~15mn from Uber’s mobility-only users in those mkts
    • Cross-platform users generate ~3x the gross bookings and profits vs single-product users, at >50% lower incremental acquisition cost vs paid channels
  • Mgmt noted that rev synergy assumptions baked into the $1.2bn target are “very conservative” relative to the cross-platform oppty

• Uber will leverage Delivery Hero’s strength in advertising to build out its offering
  • Delivery Hero has built out its ad biz to ~3% of GMV, which is higher than Uber’s current ad penetration as a % of gross bookings…mgmt views this as validation that Uber’s own ad biz has meaningful runway ahead
  • “Advertising is a very, very high-margin product. So, we’re looking forward to hearing from them as to how they are building their advertising product as well”
• Delivery Hero has also expanded “pretty aggressively” into grocery and quick commerce, getting that biz to adj. EBITDA profitability on a margin basis…Uber is “quite excited to learn from that”
• The deal will improve Uber’s competitive positioning but “recognize that’s going to take a lot of work to get there”
  • They are operating in a “super competitive” mkt, but noted that ~95%+ of competitors are pure-play mobility or pure-play delivery businesses
  • Uber’s cross-platform capabilities, global tech scale, and Uber One membership program give it the ability to “thrive in highly competitive markets” and generally grow category position while continuing to improve margins
  • “We respect our competition, and we’re always kind of paranoid about them”

• Uber will retain Delivery Hero’s HQ and keep its workforce in Berlin until at least 2029 + the Co earmarked €2bn for investments in Germany over the next 5 yrs

• Delivery Hero mgmt and Supervisory Boards “unanimously” welcome and support the transaction and intend to recommend that shareholders tender into the offer
  • Uber already holds a direct stake of 24.99% in Delivery Hero and has exposure to an additional 11.84% through financial contracts
  • Prosus, Delivery Hero’s 2nd-largest shareholder, also said it would sell its remaining 16.8% stake in the Co as part of the takeover

Source: Uber

AI news continues to move at an incredibly fast pace, with seemingly every week bringing new model launches, product announcements, funding rounds, and other major developments. Late last week, a lawsuit was added to that mix as Apple sued OpenAI, accusing the Co of stealing trade secrets related to its upcoming AI products. The first of those products is expected to be a movable, screenless AI companion for the home that becomes increasingly personalized over time and proactively assists users based on their habits and surroundings.

Apple was not the only one questioning OpenAI this week. eMarketer cast doubt on the Co’s long-term advertising projections, estimating the entire US market for standalone chatbot advertising will reach just $5.4bn by 2030, well below OpenAI’s goal of generating $100bn in annual ad revenue over the same period. Elsewhere, Google has delayed the launch of Gemini 3.5 as it works to improve the model’s coding capabilities, while over in China, Apple Intelligence finally received regulatory approval and DeepSeek continues to build momentum with annualized rev touching $500mn as it reportedly prepares to go down the IPO route.

See below for what we thought were the most incremental need-to-know updates in the AI space.

Apple Sues OpenAI, Accusing It Of Stealing Trade Secrets To Create Its Upcoming AI Gadgets (link/link/link)

• Late last week (Friday), Apple filed a lawsuit against OpenAI in which it alleged that its former employees have stolen trade secrets “for the benefit of OpenAI”
  • Apple alleges that OpenAI asked former Apple employees, and even prospective recruits, to bring information about unreleased products
  • Apple also claims that OpenAI instructed hires on how to evade its security procedures

• Apple says it first raised concerns with OpenAI directly in February, asking the Co to investigate and address the issue; OpenAI, however, never responded

• What was OpenAI’s response to the lawsuit?
  • On Friday, OpenAI issued an initial response…rejecting the accusation and saying it had no interest in trade secrets belonging to other Cos and was focused on building its own new technology
  • Then in a statement on Tuesday… “While we take these allegations seriously, we’re not aware of any evidence that this complaint has merit…we believe in fair competition and allowing people the freedom to work wherever they choose, and we’re focused on building innovative technology that empowers people everywhere”

• As part of this process, Apple has reportedly sent legal preservation letters to ~40 of its former employees now working at OpenAI
  • A preservation letter is a formal written notice sent to a person or organization telling them to preserve documents, records, and other evidence that may be relevant to a legal dispute

• For context, OpenAI has hired more than 400 other people from Apple, according to the lawsuit
  • In some instances, OpenAI poached so heavily from engineering groups, particularly within the iPhone product design organization, that Apple was forced to rebuild parts of those teams

• Apple is seeking monetary damages and an order requiring OpenAI to halt the alleged conduct and destroy any proprietary materials
  • That said, legal remedies could take months or years to materialize

On The Back Of The Suit…Reports Are Speculating That OpenAI’s First Device Will Be A Movable, Screenless Speaker Built As An AI Companion (link)

• In the immediate aftermath of the Apple filing, OpenAI reportedly believe it is on track to announce its first hardware product this yr and release it in 2027 BUT that could still change as OpenAI digests Apple’s claims

• The product (which is still under development) is meant to serve as a humanlike AI companion that lives in the home
  • It will help control smart-home appliances, play media, answer questions, respond to messages and tap into the range of capabilities offered by OpenAI’s ChatGPT
  • The machine is believed to be able to draw on personal information, such as emails, to better understand its owner

• Though the new product resembles a speaker, OpenAI reportedly internally describes it as the first of its kind…a computer built for AI to help make busy people more productive
  • It includes a camera and other sensors that help it understand a user’s surroundings and context, as well as advanced AI models beyond those available on conventional smart speakers
  • OpenAI reportedly envisions the device anticipating needs, surfacing information proactively and serving as an expert on its user

• OpenAI is said to believe that the product’s defining feature will be its personality and ability to connect with users
  • The speaker incorporates mechanical elements that can move on their own, creating a sense that it is alive and not just an object responding to commands

• The device’s communication abilities will rely on a more advanced version of the ChatGPT Voice Mode, GPT-Live, that OpenAI rolled out this month, and is designed to act more like a human
  • It can listen and talk at the same time, adapt more naturally during conversations, and quickly process information

• The speaker is designed to stay in the home, though it will be easy to move around the house
  • The device is said to include a rechargeable battery, allowing it to be carried from room to room throughout the day

-> Sonos’ stock fell -9.3% on the day of the report

eMarketer Says OpenAI’s Ad Targets Are Way Ahead Of Where The Market Actually Is (link/link)

• OpenAI has projected that ChatGPT ads could generate $2.5bn in rev this year and $100bn annually by 2030…

• …BUT eMarketer is far more skeptical, estimating that the entire US mkt for standalone chatbot advertising will generate <$1bn in 2026 and grow to just $5.41bn by 2030

• More broadly, total AI ad spending (not just chatbots) will more than double over the next 5 yrs from $32.03bn in 2026 -> $68.25bn in 2030
  • Even if OpenAI captured every advertising dollar in that mkt, it would still be far below the Co’s stated goal

• While AI search is growing quickly, most AI ad spending still depends on traditional search formats
  • Conversational search ads (i.e., ads inside search engine-based chatbots like Google AI Mode) are scaling faster than ads on standalone chatbots (i.e., ChatGPT or Gemini), which is expected to continue to face pressure on inventory, pricing, and advertiser returns

• Most AI advertising spending is expected to come from ads placed alongside AI-generated content rather than inside AI chatbots or conversations
  • 80%+ of AI ad spend in 2026 will come from placements such as search ads shown next to Google AI Overviews

Source: eMarketer

A Slowdown In The Model Race…Google’s Gemini 3.5 Is Behind Schedule

• Google delays Gemini 3.5 Pro AI model release (link/link)
  • Google is reportedly months behind schedule: The model was due to be released in June, ⁠Alphabet CEO Sundar Pichai had said during Google’s annual I/O developer ​conference in May
  • The big bottleneck? It’s coding abilities: Engineers are reportedly still working to close a coding performance gap against competitors like OpenAI and Anthropic
    • Both OpenAI and Meta recently released new models that further outpace Google’s current offerings in AI for writing code
  • Organizational sprawl is said to be a root cause: Separate teams within DeepMind, Cloud, Android, and Search are each building AI coding tools in parallel, resulting in duplicated work and slower decision-making

-> Google ended the day down -4.4% on the back of the report

Over In China…Apple Intelligence Finally Gets Approval + DeepSeek Momentum Continues To Accelerate

• Apple receives long-awaited govt approval to roll out Apple Intelligence in China, with Alibaba and Baidu as partners (link/link)
  • Apple intro’d its AI features 2 yrs ago, but the features have been stuck in the Chinese approval process + development required Apple to work with outside partners, which in Apple’s case was Alibaba and Baidu
    • China requires companies ​to register LLMs and genAI svs ⁠with regulators before making them available to the public
  • The Cyberspace Administration of China included Apple’s genAI on a list of newly approved providers, alongside recent offerings from local Cos like Huawei and Xiaomi
  • The version of Apple Intelligence for the China mkt was modified to include Alibaba technology as a filter of sorts that works with the Chinese govt to approve on-going LLM updates
    • Alibaba said in a ‌statement ⁠that its Qwen model will be integrated into Apple Intelligence across Apple’s iPhone, iPad, Mac and Vision Pro operating systems in China
  • The US version of the features taps into svs from Google and OpenAI for search, while the Chinese version works with Baidu

-> The news sent Alibaba’s ADRs up as much as +7.9% after mkts opened in New York, while Baidu’s ADRs gained as much as 4%

• DeepSeek’s annualized rev reportedly reached b/w $400-500mn, as the Co reportedly looks to raise a new round ahead of going public (link/link/link)
  • The growth has been driven largely by sales of cloud-based access to its models through APIs
  • This comes as the Co seeks to raise its second funding round of 50bn yuan, or ~$7.4bn, at a valn of 500bn yuan, or ~$74bn
    • That would value DeepSeek at ~148x its annualized rev
    • Comes just weeks after the Co closed a round that raised $7bn at a ~$50bn valuation
  • DeepSeek has reportedly maintained gross margins of 70% to 80% on access to its V4 flagship model despite charging less than OpenAI and Anthropic
    • This is because DeepSeek manages to keep the cost of running its models low due to improvements in its computing infrastructure that allow it to process more AI queries using fewer chips
  • On the road to going public? The Co is said to have hired investment banks to help it prepare an IPO in Shanghai’s tech-focused Star Market; The Co reportedly aims to file the listing application this yr and go public next yr

We have been getting increasingly concerned about the growing political and consumer push-back and back-lash regarding the massive scale and pace of data center builds given impacts to power grids and water resources, among other complaints. This week brought with it a major update on this front given New York will now be the 1st state to pause permitting for new large-scale data centers until new policy guardrails are put in place. This headlined at the same time that news hit that Meta nearly doubled its planned investment in its Louisiana data center project from $27bn to more than $50bn. To redirect concerns, the Co highlighted the economic benefits the project is bringing to the region but big picture, we would expect to see this heightened political and consumer sensitivity persist

At the same time, reported multi-billion dollar compute leasing discussions between Meta and Anthropic, as well as SpaceX and the U.S. Department of Defense, underscore that demand for AI compute is showing no signs of slowing down anytime soon. Morgan Stanley also raised its capex forecasts for the hyperscalers, arguing that the next phase of the AI race will be defined less by who builds the most capacity and more by who can monetize it.

See below for more of what we found most important on this front this week.

New York Imposes The Country’s First Statewide Moratorium On Data Centers…But Not Without Presidential Backlash (link/link/link)

• The order will pause state permitting for new large data centers (w/ capacities greater than 50 megawatts) for up to one yr + direct state regulators to create standards that address environmental impacts, energy demand, water usage and other factors

• The moratorium will be lifted once the govt is able to execute on a policy that sees AI Cos contributing some of the growing energy bills
  • “We’re going to explore having hyperscale data centers pay into a larger fund to support our grid statewide. We expect this process, which we already launched, to be completed within the year. Once this policy is in place, the moratorium will be reviewed and lifted,” NY Gov. Kathy Hochul said

• Other key focuses of the plan –
  • Make sure data centers are only built “in places that want them,” so they will not be exempt from local zoning and approvals
  • Include labor issues in discussions with data centers hoping to build in New York…”projects can include wage standards and labor agreements, and prioritize local hiring”
  • Include a proposal to end tax subsidies for data centers

• “As data center development threatens to hike up utility bills, deplete our natural resources, and create uncertainty for New Yorkers, it’s my responsibility to take action and lead,” Hochul said in a statement

• But not everyone is in support…President Trump in a Truth Social post said that the state should change its policy “IMMEDIATELY” …
  • “One of the biggest Driving Forces in the Future for Jobs, are Data Centers,” he wrote. “They are big, strong, bold, and Money Machines for the State in which they are built… New York State has made a terrible decision”

• …to which Hochul responded with her own post on X – “We hit pause because the communities powering AI should share in its success. Maybe that’s a novel concept in Washington. We call it doing our job”

-> As a reminder, Sen. Bernie Sanders (I-Vt.) and Rep. Alexandria Ocasio-Cortez (D-NY) brough the idea of a data center moratorium into the national spotlight earlier this year with legislation that would have paused new projects nationwide; The Sanders-AOC proposal went even further and would lift the ban only after Congress passed sweeping AI safety legislation covering issues from civil rights to consumer protection (link)

• Moratoriums have been proposed in at least a dozen states but have not gotten far, though some counties and municipalities have imposed their own temporary bans
  • Earlier this year, Maine seemed poised to establish a similar moratorium, but the measure was vetoed by the state’s Democratic Gov. Janet Mills because it would have blocked a proposed data center in a town that has struggled following the closure of a local mill

As New York Slows Data Center Development, Meta Accelerates Its Louisiana Buildout

• Meta is scaling its Northeast Louisiana data center project to 5 GW of compute capacity, with costs now exceeding $50bn (link/link/link)
  • That’s higher than the $27bn that was first shared in Oct, when Meta and Blue Owl Capital formed a JV to help with the build-out and mgmt of the facility, originally planned as a 2 GW data center
  • The site will encompass more than 3,200 acres, which is more than 4x larger than New York City’s Central Park

• Meta highlights “life-changing” returns for teachers, local bizs and students in the area –
  • Since breaking ground in Dec 2024, local Louisiana businesses have received $1.6bn+ ‌in ⁠contracts from Meta
  • With this expansion, Meta plans to invest $1bn+ in local infrastructure improvements, including roads, water and wastewater systems.
  • Teachers in Richland Parish recently received annual bonuses of up to $50,000, a +400% ⁠increase from the previous yr, thanks to incr’d tax revs from the Co’s data center project.
  • Also the Co is donating $5mn to Louisiana Delta Community College to create scholarships for residents training for data center jobs

• Meta pays the full costs of the energy, water, and related infrastructure the data center uses so consumers aren’t paying the cost
• Meta currently has 32 data centers ​across the globe in operation or under construction, 28 of which are in the US

-> Meta’s shares fell -1.9% on the day of announcement 

Demand for AI Compute Continues to Accelerate as Meta and SpaceX Reportedly Explore Multi-Billion Dollar Leasing Deals

• Meta is said to be in talks to lease $10bn in computing power to Anthropic, per the New York Times (link/link)
  • Anthropic is reportedly in early discussions to lease computing power from Meta, potentially worth as much as ~$10bn over 2 yrs
  • Anthropic, which reportedly proposed the deal in June, would pay Meta in monthly increments over the 2-yr term and the Cos would be allowed to opt out of any agreement early
  • For Meta, a deal would open a new line of biz for the Co, though the talks are in early stages and may not result in an agreement
  • Anthropic made a similar deal w/ SpaceX back in May… under that agreement, Anthropic is paying SpaceX $45bn over 3 yrs (or $1.25bn/mo) for computing power; The deal included similar provisions that let either Co exit the agreement early

• SpaceX is reportedly in talks w/ the Defense Department about providing computing power worth billions of dollars for running AI models (link/link)
  • The talks are reportedly centered around providing the agency with access to data-center capacity worth billions of dollars for running AI models
  • The talks are ongoing and could fall apart
  • The deal is similar to previous agreements SpaceX has reached with Google and Anthropic
  • A Defense Dept agreement would mark the latest deal between SpaceX and the Pentagon, which relies on the Co for launching rockets and managing satellites for communications and missile tracking

Morgan Stanley Raises Hyperscaler CapEx Forecasts as the AI Race Shifts From Building Compute to Monetizing Capacity (link/link)

• Morgan Stanley raised its CapEx forecasts for the five major hyperscale cloud operators, increasing its 2027 estimate by +9% to ~$1.2 trillion and its 2028 estimate by +10% to ~$1.4 trillion
  • Includes Meta, Amazon, Microsoft, Google, and SpaceX

• What drove the upward revision? Soaring hardware costs and data center construction delays
  • Estimates that overall GPU-related costs have risen by ~20%
  • Data centers can now take up to three years to become operational + community opposition and political uncertainty ahead of the 2028 U.S. election is prompting hyperscalers to break ground early, which is accelerating the front-loading of CapEx

• The available computing power of the five major cloud hyperscalers is predicted to grow from ~30 GW in 2025 to ~116.6 GW in 2028

• The AI investment story is evolving…
  • In 2025, the market was most concerned with who could secure GPUs
  • In 2026, it began to focus on who could build data centers
  • By 2027 to 2028, what will truly determine valuations is who can sell that capacity, converting it into advertising, cloud services, API, and subscription revenue

• Select Morgan Stanley company-specific callouts / views –
  • Meta has the most underappreciated AI monetization story
    • investors are focusing too heavily on Meta’s CapEx and underestimating its ability to monetize AI through advertising and lower-priced model APIs
    • Even a small allocation of computing capacity to APIs could create a meaningful new rev stream
  • Amazon AWS has the clearest near-term monetization path
    • Strong cloud demand, a growing backlog, and large contracts from AI companies could support rev growth
    • Higher AWS revenue may help offset the depreciation and capital costs tied to increased spending
  • Google has strong technology but near-term capacity constraints
    • Google is currently in a state of tight computing capacity, evidenced by its recent computing power lease agreement with SpaceX
    • The computing bottleneck could drag down its near-term rev growth or the pace of new product launches, a risk that is relatively higher than for Meta and Amazon
    • Also, the release schedule for Google’s next-generation model, Gemini 3.5, has already fallen behind the company’s original plans outlined at its I/O developer conference

The studio and media entertainment sector saw a flurry of activity this week spanning M&A speculation, box office milestones, major franchise marketing moves, policy lobbying, and analyst calls on streaming strategy. Details are below.

More Studio M&A?

• Lionsgate is reportedly exploring a sale, w/ takeover interest from Bolloré and Banijay (link/link): Lionsgate apparently is working w/ an investment bank to evaluate inbound approaches
  • Background on these interested parties –
    • France’s Bolloré Group’s interest is driven by a desire to bolster the production capabilities of Canal+, the pay-TV Co in which it holds a controlling interest
    • Banijay Group (TV production Co behind “Big Brother” and “Survivor”) earlier this year completed its merger with All3Media and is reported to be among suitors that have considered a bid
  • Netflix last month denied acquisition interest
  • It is good timing w/ Lionsgate’s Michael hitting $1bn in global box office (see below), which is its 1^st
    • And surpassed The Hunger Games: Catching Fire ($865mn)

-> Lionsgate shrs rallied as much as +10% on the back of the news and has skyrocketed by +52% YTD

This Week Was Marked By Both Booms & Busts In The Global Box Office

• Michael became the first $1bn biopic in box office history (link/link/link/link): Michael global box office reached $371.8mn domestically and $629.8mn overseas after 12 weekends in release
  • The film surpassed other biopics…
    • Bohemian Rhapsody ($911mn)
    • Oppenheimer ($975.8mn)
  • It is the 2^nd film to reach the $1bn mark this year (1^st being The Super Mario Galaxy Movie at $1.008bn)
    • Toy Story 5, currently at $879mn, is expected to join that club soon

• But Disney’s live action “Moana” opened more softly than hoped (link/link)
  • Moana opened to $95.1mn worldwide ($52mn international, $43.1mn domestic)
    • It was described as underperforming with “very low openings”
    • Releasing too close to Moana 2 (Thanksgiving 2024) was cited as a culprit
    • Deadline ests the film is poised to lose between $100–$125mn (assuming the movie hits $250mn global box office)

• Key films to watch…(link/link)
  • Disney’s Avengers: Doomsday…domestic premium tickets to go on sale July 20^th
    • This is 5 months ahead of its December 18 theatrical release
    • Tickets will be for Infinity Vision theaters (Disney’s premium large-format certification created in response to Warner Bros. locking down IMAX screens for Dune: Part Three, which opens the same day)

• Amazon MGM Studio’s Spaceballs: The New One’s theatrical release was confirmed to be April 2027: This would be the 40^th anniversary of the original 1987 film

A Push For Disney To Exit Streaming

• Wells Fargo this week moved Disney’s stock when it argued that the Co should exit the DTC streaming business and return to a pure content-licensing model (link/link/link)
  • Why exit streaming? Wells Fargo analysts argues that Disney is structurally unable to compete with high-volume streamers like Netflix and YouTube and that the rest of their businesses wouldn’t suffer if their library was on a competing global streamer
  • They est that Disney could generate $15bn+ in annual licensing revenue if it focused purely on content vs. distribution
    • Licensing could add ~10% to EPS by fiscal 2028
    • A shift of this nature could boost the stock by ~40%
  • In the meantime, while the analyst maintained their Overweight rating, but cut their PT from $146->$125/shr

-> Disney shares are down by ~50% over the past 5 yrs

Following last week’s US vs Belgium Round of 16 blockbuster viewership (~33mn English language viewers on Fox’s telecast, the most watched soccer telecast in US history), the World Cup ratings machine kept on giving this week with Fox’s back-to-back record semifinal telecasts (Argentina vs England at 15.1mn, Spain vs France at 11.5mn). Both were more than double the prior English language semifinal record. We are looking forward to this Sunday’s Argentina vs. Spain final match!

An interesting, related survey published by Samsung Ads this week shows that streaming has overtaken cable for World Cup viewing for the 1st time at a major sports event (56% vs 37%, respectively) which is important as leagues and sports rights holders assess the best distribution channels to maximize value. Another theme in sports this week was what seems like forever escalating value of sports teams. This past Saturday, a Khosla-led group agreed to pay $9.6bn for the Seahawks, which is the highest price ever for an NFL franchise.  Rogers also is now taking full ownership of MLSE at an implied $17.4bn valuation, up 39% from just last year. With that said, not all sports assets are inflating. The Bundesliga’s new US deal with Versant came in at $20mn/yr per press reports, which is a 33% haircut from what ESPN was paying before. The league is swapping dollars for distribution by moving to free, ad-supported streaming on Fandango in a bid to “expand accessibility.”

Overall, live sports remains the most valuable content in media, but there is a gap between tier 1 and tier 2 properties. Below is more detail on all of these developments this week.

The World Cup Continues To Break Viewership Records! & WC Streaming Viewing Surpasses Cable…

• FOX aired the 2 largest English-language World Cup semifinal telecasts in US history in consecutive days (link/link)
  • Wednesday – Argentina vs England: Avg’d 1mn viewers on Fox
    • It was the most-watched FIFA Men’s World Cup semifinal ever in English-language US TV
    • Peak viewership was 22.2mn from 4:45–5 p.m. ET
  • Tuesday – Spain vs France: Avg’d 5mn viewers on Fox
    • It held the semifinal record for ~24 hours before Argentina/England topped it
  • For context…the prior English-language semifinal record was France vs Morocco in Dec 2022 at ~6.6mn on Fox hence both 2026 semis more than doubled that
    • But viewership for Weds’ match up was still a far cry from Fox’s US vs Belgium Round of 16 at ~33mn viewers, which is the most-watched soccer telecast in US history

• A Samsung Ads survey found streaming topped cable for World Cup viewing for the 1st time at a major sports event: 56% streaming vs. 37% cable (link)
  • Of streamers:
    • 42% used an existing subscription
    • 14% signed up for a new svs specifically for the tournament
  • Among new sign-ups – most joined via free trial or lapsed-account reactivation, not fresh paid plans
    • YouTube TV led, accounting for 35%
    • Peacock accounted for 34%
    • Hulu + Live TV accounted for 31%
  • For actual watch time during the tournament –
    • Peacock accounted for 41%
    • Hulu + Live TV accounted for 34%
    • YouTube TV accounted for 29%
  • Smart TVs were the dominant device at 61%; streaming sticks (Roku, Fire Stick) were a distant second at 17%
  • The churn risk…
    • 56% of new subscribers say they plan to cancel once the tournament ends
    • 44% intend to stay

NFL Franchise Values Hit A New Record!

• On Saturday, a Khosla-led group agreed to buy the Seahawks for a reported $9.612bn which would be the highest price ever for an NFL franchise, surpassing the Commander’s 2023 sales of $6.05bn in 2023 (link/link)
  • This also represents the 2^nd largest sports franchise transition in N. America (#1 is the $10bn LA Lakers deal in 2025)
  • Paul Allen purchased the Seahawks in 1997 for only $194mn and his estate will give all sale proceeds to charity

• This follows the week prior when Rogers Communications annc’d that it is buying the remaining 25% ownership stake in Maple Leaf Sports & Entertainment (MLSE) from Kilmer Sports as it previously telegraphed it would do; Rogers will now own 100%…the deal size is C$4.35bn (link/link)
  • What does MLSE own? The NHL’s Maple Leafs, NBA’s Raptors, MLS’ Toronto FC, Toronto Argonauts
  • Rogers other sports ownerships/partnerships:
    • Ownerships in the Toronto Blue Jays, Rogers Centre & Sportsnet (the #1 sports media brand in Canada)
    • Strategic partnerships w/ the Vancouver Canucks, Edmonton Oilers, Calgary Flames, the NHL, the NBA, MLB and Live Nation
  • MLSE’s value has materially gone up:
    • Last year, Rogers purchased BCE’s 37.5% stake in MLSE for C$4.7bn (took Rogers ownership to 75%)
    • This deal w/ Kilmer implies a total value for MLSE of C$17.4bn, a 39% increase from the $12.5bn valn implied w/ Rogers’ acq of BCE’s stake
  • The transaction should close in Q4

But Sports Rights Don’t Only Go Up…

• Versant (NBCU spinoff) this week acq’d exclusive English-language US rights to the Bundesliga for 5 seasons (2026/27–2030/31) (link/link/link)
  • The total $s paid declined vs the last deal
    • Value = $100mn total ($20mn/yr), down from the $30mn/yr ESPN was paying under the prior 6-season deal
  • What’s included?
    • 30+ matches/season on linear USA Network
    • The remaining 250+ matches streamed free (ad-supported) on Fandango, which has never carried live sports before
  • Why accept a lower price? The Bundesliga accepted a ~33% annual fee haircut in exchange for broader reach, moving from behind the ESPN+ paywall to a free AVOD platform reaching 80mn+ US homes
    • The priority was “expanding accessibility” while delivering “strong financial results.”
  • For Versant, the deal extends USA Network’s growing live sports portfolio (already has EPL, WNBA, college sports, golf) and positions Fandango as a potential live-sports AVOD competitor

It was a rough week for US wireless/cable incumbents, w/ legacy plan shake-ups, another round of headcount cuts, and some Wall Street analysts voicing concerns about SpaceX’s Starlink ambitions compounding sector pressure. A lot of negativity is certainly priced into the stocks, especially for cable. Earnings reports for the Connectivity sector next week will be the next potential catalyst for the group and all of the below will be in focus.

-> For the week, CHTR was up +0.5%, CMCSA +0.9%, TMUS +2.6%, AT&T +3.2%, VZ +3.5%, while SPCX was down -14.7%

• Following speculation earlier in the week that job cuts were on the way (per Barron’s, confirmed by an employee familiar w/ the plans), Verizon officially annc’d a 3rd wave of layoffs on Thursday (link/link/link/link):
  • Cuts impact a total of ~3k employees –
    • ~500 corporate employees and selling 274 company-owned retail stores to franchise operators (~2,500 retail employees affected)
    • The store sales become effective Aug. 16, leaving Verizon w/ ~1,000 Co-owned stores
    • This follows the sale of 179 corporate-owned stores to franchise operators in Nov 2025
  • The cuts are reportedly part of CEO Dan Schulman’s target to slash $5bn in 2026 opex
    • But previous estimates suggested that 8-10k employee cuts might be needed to hit the $5bn target
    • Other areas of opex optimization beyond headcount include: vendor optimization, software streamlining, real estate footprint reduction, and AI deployment in customer service
  • These cuts follow the record 13k-employee redundancy in Nov 2025 (~13% of workforce) and a smaller round of several hundred cuts in May 2026
    • VZ entered 2026 w/ ~89.9k employees

• Wall Street’s Bernstein analyst (Laurent Yoon) weighed on Telco/Cable this week with PT cuts across the group on Monday, citing valn risk from SpaceX’s Starlink ambition (link/link/link/link)
  • What were the PT changes?
    • VZ PT from $49->$44 (Mkt Perform)
    • AT&T PT from $30->$25 (Outperform)
    • T-Mobile PT from $245->$220 (Mkt Perform)
    • Charter PT from $210->$170 (Mkt Perform)
    • Comcast PT from $32->$28 (Mkt Perform)
  • This came on the back of recent market concerns about Starlink Mobile potentially pursuing a DTC Starlink mobile service and exploring building its own terrestrial wireless network in the U.S
    • SpaceX holds 65 MHz of exclusive-use, nationwide mid-band spectrum (AWS-4, AWS-3, H-Block) via the FCC-approved $17bn EchoStar acq (May 2026); Licenses fully transfer in late 2027
    • BUT all 3 major carriers have publicly or reportedly said they would decline to offer SpaceX an MVNO deal
  • Bernstein’s view: Starlink is unlikely to materially disrupt incumbents in the near-term, but adds another competitor to an already saturated market and introduces persistent uncertainty that compresses multiples
  • Oher Wall Street views:
    • MoffetNathanson’s Craig Moffett argues that physics prevents meaningful suburban penetration
    • Wolfe Research’s Peter Supino called Starlink “a comet bearing down on broadband incumbents”
    • Morgan Stanley’s Sean Diffley said “we believe the perceived risk of Starlink Mobile disrupting the US Wireless industry is greater than the actual risk” and would lean more on partnerships vs building its own network.

-> This follows Wells Fargo’s Steven Cahall last week initiating on the US wireless industry with a cautious outlook, flagging Starlink as a growing competitive threat to FWA and postpaid account growth

• T-Mobile makes changes to some legacy plans (link/link/link)
  • What plans are being retired?
    • Simple Choice
    • ONE/ONE Plus
    • Magenta
    • And grandfathered Sprint plans
  • What plans are these subscribers being transitioned into?
    • Experience-branded rate plans
    • Users will keep their current benefits but will also gain addtl perks & improved service experiences
    • T-Mobile is also offering a 5-year price guarantee on the new plans
  • But the push back is that it means higher prices for some of these long-time loyal subscribers
    • “A very big portion” of customers will NOT see a price change
    • But others will face increases of ~$4–6/line/month (more or less) but still pay below “rack rate” for the equivalent current plan
  • For T-Mob, this move simplifies the back-end, which should accelerate feature rollouts and reduce support costs over time
    • It consolidates ~1,100 legacy billing codes down to <100
  • Separately, customers are reporting online that the T-Mobile Kickback program, which provides credit if you use less than 2GB of data in a month, is being retired along with the plans

• Google is making moves into Pinterest’s territory (link/link)
  • The Co unveiled a significant redesign of Google Images on its 25th anniversary, which transforms the product from a basic search grid into a Pinterest-style discovery feed w/ personalized “For You” galleries, saveable Collections (functionally identical to Pinterest boards), and continuous-scroll browsing
    • The update also embeds AI image generation directly into Search via Google’s Nano Banana model so users can create custom visuals from text prompts w/o leaving the ecosystem (a play to keep sessions away from chatbots)
  • Timing: The update is rolling out over the coming weeks on desktop in the US in English

-> PINS was flat on the back of the news and ended the week up +3.0%  

• Publicis posted better than expected results this week and raised the low end of its 2026 revenue guidance (FactSet/link)
  • Q2 net revenue organic growth BEAT
    • Accelerated to +4.8% y/y (vs. +4.5% y/y in Q1) and beat cons +4.5%
  • Co RAISED the low end of the FY26 organic growth guidance
    • To +4.5–5%
    • From +4–5% previously (Publicis Groupe; Adweek)
  • Strength came from its AI-powered marketing services (Epsilon, media, creative – now accounting for 87% of net revenue, up +1pt from Q1)
  • Sapient (technology/consulting, ~13% of net revenue) continues to be a drag
    • Fell mid-single digits in Q2, reflecting industry-wide delays in client transformation spending amid macro uncertainty
    • Mgmt said no major recovery is assumed in H2, but the unit is stabilizing sequentially
  • Q2 op margins at 17.5% were well ahead of cons 16.5%
    • And for 2026, the Co maintained a “slight improvement” vs 18.2% in 2025 (despite high levels of investment)
  • The Co confirmed its 2027 and 2028 objectives for avg net revenue and headline EPS growth at constant currency of at least +7% to +8% and +8% to +10% per year, respectively

-> Publicis fell -3.9% post its print and is down -3.1% YTD  

• StubHub shares come under pressure from the DC resale cap (link/link/link)
  • Washington, DC Council unanimously passed the RESALE Act on Tuesday, introducing a 10% cap on ticket resale prices above face value for concerts (the strictest price cap yet)
    • The act also bans speculative ticket sales (listing tix you don’t possess), requires full price transparency, and mandates registration for sellers moving 50+ tickets/year
    • Next steps: The act now heads to Mayor Muriel Bowser for approval and is expected to go into effect on January 1, 2027
  • Citi’s analyst flagged significant EBITDA risk: They est’d that it could shave about $95mn off StubHub’s 2027 EBITDA if one‑fifth of ticket volume is subject to the cap
  • Could other states follow? State legislators in NY and CA, as example, are considering price cap legislation as well

-> StubHub fell -11.3% on the day of and is down -31.6% YTD