February 2026 LLM Market Share Snapshot

February 2026 was a watershed month for large language models: growth is still explosive, but power is clearly consolidating around a handful of consumer chatbots and enterprise platforms.

Market size and momentum

The wider LLM and generative AI market continues to expand at a double‑digit clip, driven by enterprise adoption and AI‑native products. Analysts now peg the global LLM market in the mid–single digit billions, with forecasts pushing it toward tens of billions by the early 2030s. Transformers remain the dominant underlying technology, controlling a large share of generative AI revenue across language, vision, and multimodal workloads.

February 2026: Consumer LLM usage

On the consumer side, February 2026 confirmed that we’re in an oligopoly, not a monopoly. ChatGPT remains the primary entry point for most users, but its share is steadily eroding as Gemini, Copilot, and a long tail of vertical assistants mature.These ranges blend app‑store share, web traffic, and AI search data; they are best interpreted as directional rather than precise.

Gemini’s February Breakout

If February had a single headline, it was Gemini’s acceleration. In AI search share, Gemini has broken into the high teens to around 20%, up significantly year‑over‑year. Mobile usage shows a similar pattern as Gemini benefits from default placement and deep integration across Android and Google Workspace. For product and growth teams, this is a clear demonstration that distribution and ecosystem lock‑in can chip away at an early‑mover advantage.

ChatGPT: From Monopoly to Anchor Tenant

OpenAI’s ChatGPT is still the anchor tenant of the LLM ecosystem, but the days of near‑70% dominance are over. Its share has fallen into the ~50% range even as absolute usage continues to grow. In AI search experiences, ChatGPT‑based offerings still command a majority, but that number has started to plateau. The implication: we’re shifting from “everyone launches on ChatGPT” to a multi‑model reality where Gemini, Claude, Copilot, and open‑source backends coexist behind the same UI.

Second Wave: Claude, Copilot, Perplexity and Others

• Below the top two, a second wave of fast‑growing specialists and ecosystem plays is emerging.

• Microsoft Copilot sits in the low‑teens, powered by distribution inside Windows, Office, and Edge rather than pure consumer brand pull.

• Claude, while still in the single digits, is one of the fastest‑growing enterprise assistants, especially for long‑context knowledge work and development.

• Perplexity, with a few points of the AI search market, over‑indexes among power users looking for conversational search and source transparency.

Chart Candy: Solo Activities, OpenAI, SaaSpocalypse

CHART Candy: Sharing a few compelling charts from the past week covering retail, marketing, AI, and the economy. Let the visuals do the talking - swipe through the carousel to see what caught my eye.

1. Solo activities are all the rage. "Restaurant for one." I'm not ashamed. I've done it. Just don't tell anyone.

2. Despite "SaaSpocalypse," software developer employment is up considerably more than decline in computer programmers. AI driving more building thus far.

3. Instagram now reigns supreme again vs TikTok in the US based on daily active users.

4. OpenAI's ChatGPT may have well over 900 million daily users but Anthropic's (Claude) has more subscriptions for businesses.

5. Red Bull continues to dominate energy drinks and just hit another inflection point. Guess Max didn't have his Red Bull this past weekend at the Australian Grand Prix.

6. Mortgage rates are nearing 2022 levels.

7. Major tech blogs and publishers see dramatic drops in traffic due to paywalls and generative engines such as ChatGPT and Claude.

8. Greater than 50% of every shopping demographic at least "somewhat agrees" that return policies drive retailer choice.

View fullsize

View fullsize

View fullsize

View fullsize

View fullsize

View fullsize

View fullsize

View fullsize

More of the Same: AWS & Advertising Power Profit Growth Despite Rising Retail Inventories

The close of fiscal year 2025 for Amazon shows much of the same from prior quarters. A continued acceleration in the services business with AWS growing 24% year-over-year as Advertising and Third-Party Seller Services each drove double digit growth. Although the margin was difficult in international markets, the North America market margin exceeded expectations despite inventories growing a bit faster than retail sales. Here are the key highlights:

AWS Accelerates Above Expectations
Amazon Web Services (AWS) grew 24% year-over-year, outpacing consensus and adding $6.8 billion in incremental sales. AWS continues to capitalize on the industry-wide shift toward cloud and focus on AI, with Google and Microsoft also reporting healthy gains, albeit from smaller bases. The growth rate was the best AWS has seen in 13 quarters.

Marketplace and Seller Services Fuel Growth
Third-Party Seller Services delivered a second third consecutive quarter of double-digit growth. Since Amazon takes a referral fee from these transactions, strong 3P sales signal not just higher volume but greater pricing power. This business now accounts for nearly 25% of the overall net service sales.

Advertising Revenue Momentum Continues
Amazon's advertising business grew 24% this quarter, building on last quarter's 24% growth. This surge comes as Amazon steadily grabs market share from rivals like The Trade Desk, strengthening its position as a key force in digital marketing. Key partnerships with players like Netflix, Roku and Spotify should keep this key source of revenue growing for future quarters.

Online Sales and Inventory Imbalance
Core online sales rose just under 10%, nearly hitting the third straight quarter of double-digit growth. However, inventory levels climbed even faster: up 12% this quarter and 14% in Q3 and 19% in Q2. While the pace of inventory accumulation is slowing, the overhang remains. This imbalance points to elevated pricing and stepped-up vendor funding as Amazon seeks to convert heavy inventory into margin gains.

Amazon's latest results underscore its ability to CONTINUE to grow both the top and bottom line-even as the retail landscape remains fiercely competitive and inventory levels remain in focus. Although the market is spooked by the $200b in planned capital expenditures, nothing here suggest Amazon slowing anytime soon.

Is AI making employees more productive?

I know. You are sick of hearing about AI. And debating whether AI is actually moving the productivity needle. So is it making everyone more productive? It seems so.

a16z’s latest benchmarks show ARR per employee at top SaaS companies has roughly tripled since 2018, with the 90th percentile now approaching about $700K ARR per FTE.

At the same time, developer behavior is loud and clear: daily installs of AI coding assistants in VS Code keep climbing, with tools like Claude Code, Gemini, and OpenAI-powered extensions seeing sustained, growing adoption rather than a fad spike. This is daily new installs on a 30‑day moving average, not just cumulative numbers creeping up in the background.

Can we attribute all of that ARR/FTE expansion to AI? Of course not. There’s pricing, product‑market fit, GTM efficiency, and macro normalization in the mix. But it’s hard to argue AI is “not working” when the highest performers are pulling away on revenue per head, while simultaneously leaning into AI tooling across their engineering orgs.

The more interesting question for operators isn’t “Is AI overrated?” but “Why aren’t we seeing these gains yet?” In most orgs, AI is still a bolt‑on assistant, not a rewired workflow in areas like coding, support, and sales ops where the leading teams are already compounding advantages.

Bottom Line: AI isn’t a silver bullet, and mis‑implemented tools can absolutely waste time but it’s clearly becoming a force multiplier, not a drag. The gap is no longer in the tech; it’s in how aggressively and thoughtfully leaders are redesigning work around it.

Well Played Zuck

What's the one thing OpenAI and Anthropic have that Mark Zuckerberg doesn't? Subscription revenue. 📰

Meta’s latest move says more about its business model than its product roadmap.

With Manus AI, a $2B+ AI agent acquisition that was already doing significant subscription revenue, Zuckerberg didn’t just buy technology—he bought a proven SaaS engine he can plug directly into Instagram, Facebook, WhatsApp, and Meta AI and own a subscription model.

Now Meta is testing premium subscriptions across all its flagship apps: “pro” AI features, advanced creation tools, workflow automation, and power-user utilities on top of the existing free social graph.

Manus will be integrated into those experiences while still sold as a standalone subscription for businesses, giving Meta both B2C and B2B recurring revenue out of the same agent stack.

Why this matters strategically:

• OpenAI and Anthropic were born with subscription and API revenue at the core (ChatGPT Plus, enterprise seats, API usage), not advertising.

• Meta was born as an advertising machine and is now trying to retrofit a subscription business on top of billions of users.

• High-compute AI features (agents, video generation, automation) are too expensive to fund purely with ads - paywalls and tiers are the cleanest way to make that unit economics work at scale.

  So Manus is Meta’s shortcut: instead of slowly building a subscription culture from scratch, Zuckerberg is using a native “AI agent SaaS” business to:

1. Monetize beyond impressions and clicks

2. Tie revenue directly to outcomes and productivity, not just attention

3. Narrow the strategic gap with OpenAI and Anthropic on recurring AI revenue

If you zoom out, this is Meta’s first serious attempt to answer a simple question: in an AI-first world where assistants, not feeds, are the primary interface… how do you make money when you don’t own the subscription relationship?

Zuckerberg’s answer: buy the agent, wire it into the feed, and charge for the work it does across the entire ecosystem. Well played Zuck...right before earnings this coming Wednesday.

The Rip's Matt Damon on How Netflix Has Changed

OFF TOPIC on Matt Damon's The Rip on Netflix.

Matt argued streaming hasn’t just changed where we watch movies, but what gets made, how it’s written, and who makes money from it:

On Netflix specifically, Damon says the platform now asks creators to reiterate the core plot three or four times, assuming many viewers are half-watching while on their phones interacting with their phone notifications. Additionally, big action set pieces must hit in the first 5-15 minutes of the film to keep the viewer hooked and not drifting away.

That means the product is increasingly engineered around distracted attention and algorithmic engagement, not the kind of dense, layered storytelling that rewards full focus.

Damon and Ben Affleck have tried to hack this system by structuring their Netflix deal so that not only stars but below-the-line crew share in bonuses if the film overperforms on the service.

At the same time, Netflix is rapidly building an advertising business: ads are only about 3% of overall revenue today (averaging $0.40 per month per subscriber), but ad dollars are growing much faster than subscriptions and are central to its long-term strategy.

As Netflix leans harder into advertising, the incentives to design movies that maximize time spent with early hooks, constant exposition, never-let-them-drop-out storytelling only get stronger.

Damon’s comments suggest streaming platforms function more like giant ad and engagement engines with muted creative ambition. Views & Ad Dollars > Subscriptions.

Note: Netflix modified MAUs to MAVs in Calendar Year 2025.

Alexa+ Joins the Daily Active User AI Race

Alexa (under name Alexa+) EARLY ACCESS for the browser has finally launched inline with Day 1 of CES. Previously Alexa was only available via Echo devices. In reading the fine print, Alexa+ is available for non-Prime members through January 2026 and will remain free for Prime members. Access will be available for $19.99 per month as a stand-alone subscription post January 2026.

The incumbents clearly have a head start in time:

• ChatGPT - November 2022

• Perplexity - December 2022

• Claude - March 2023

• Gemini (as Bard) - March 2023

Daily Active Users (DAUs) for incumbent sites continue to grow year over year:

• ChatGPT - >220mm, +114% YoY

• Perplexity - >8mm, +185% YoY

• Claude - >3mm, +219% YoY

• Gemini - >6mm, +913%

View fullsize

View fullsize

View fullsize

View fullsize

When 1% Beats 99%: Gen AI’s Tiny Traffic, Huge Conversion, and Why Walmart Is Winning the Next Demand Moat

Gen AI referrals are still a small slice of the pie in absolute terms, but they are already punching far above their weight in impact.

How big is Gen AI vs other referrals?

Across major industries, AI platforms currently drive only about 1% of overall web traffic, meaning 99% still comes from search, social, direct, email, and other traditional sources.

In June 2025, Similarweb estimated roughly 1.1–1.13 billion AI‑driven referral visits vs about 191 billion from Google Search alone, so Google is still sending ~170x more clicks than AI platforms.

Even at ~1% of traffic, AI referrals convert at materially higher rates, with Similarweb citing ~7% conversion to transactional sites, outpacing traditional search and most other referrers.

​Who is winning and losing?

Walmart. Walmart’s growing share of Gen AI referrals in the chart matters: it is winning a small but high‑intent channel that is compounding quickly, while Amazon’s resistance to AI scraping risks trading today’s data moat for tomorrow’s demand moat.

Amazon, meanwhile, is ceding relative share in the stack as more of the incremental AI-driven clicks flow to competitors like Walmart, Temu and AliExpress.

Throughout 2025 Amazon has steadily expanded its robots.txt blocklist to keep AI crawlers from Meta, Google, OpenAI, Anthropic, Perplexity and others off its site, explicitly trying to stop models from training on or browsing its eCommerce data.

That might protect short‑term data moats, but it also means many shopping agents, comparison copilots and answer engines see less Amazon inventory, fewer offers and weaker price signals when they build their recommendations.

​Walmart’s “open” bet on AI discovery

Walmart has not taken the same aggressive, public stance against AI crawlers in robots.txt and has become a favorite dataset for third‑party tools, scrapers and analytics platforms that help power AI‑driven shopping and GEO (Generative Engine Optimization).

By allowing more structured access to its catalog and pricing signals (within its own anti‑bot guardrails), Walmart is effectively treating Gen AI platforms like the next generation of search—places where being “most visible” beats being “most protective.”

Does Amazon care?

In my view, not yet. Gen AI referrals convert at a very high rate, but the absolute volume is still tiny and remains a rounding error next to legacy channels like search and social. Even with triple‑digit year‑over‑year growth, AI traffic still pales in comparison to Google, which sends roughly two orders of magnitude more clicks than AI platforms today. When will they care? Tough to say but the deals to spend $38 billion on AWS and an investment and invest. $10 billion in OpenAI will certainly grease the skids. At this point, Walmart has more to gain than Amazon does.

2026 Commerce, Retail Media & AI Predictions

As we close out 2025, it’s time to put some chips on the table.

Here are a few bold 2026 predictions for commerce, retail media, and generative AI — some controversial, some grounded in very real data and behavior shifts already underway.

• Agentic Commerce Remains Overhyped

  • Major retail platforms make their money by owning discovery, selling advertising and monetizing first‑party data, so truly open agents that roam catalogs and transact autonomously impact their economic model.

  • AI Browsers have yet to take off and lack mobile. Perplexity’s Comet and OpenAI’s Atlas have launched with limited fanfare. Atlassian’s Dia, Google’s Project Mariner and other experiments will remain niche.

  • When purchases aren’t obvious (travel, electronics, fashion), consumers still want control, brand cues, and post‑purchase accountability; offloading everything to an agent is convenient but not trustworthy.

  • The real power will sit with a handful of platform-level agents (Amazon’s Rufus, Walmart’s Spark, TikTok, Apple, Google), turning most brand-side agents into glorified feed optimizers rather than true customer interfaces.

• TikTok Shop Exceeds $100 Billion in GMV Surpassing eBay and Temu

  • TikTok’s growth rate and adoption provide an audience to cement the media destination as a commerce destination exceeding both eBay and Temu.

  • Although TikTok Shop is still skewed to trend‑driven goods and merchants tend to be reseller bundles, brands continue to take the platform more seriously as the new Bytedance and US investor ownership model has set in.

  • TikTok isn’t just another marketplace. It’s where discovery, content, and conversion collapse into a single surface — and that’s the threat every other platform is scrambling to answer.

• Brands Will Seek Tariff Refunds

  • A major Supreme Court case (Learning Resources v. Trump) could invalidate broad Trump-era tariffs imposed under the International Emergency Economic Powers Act (IEEPA); if that happens, importers may be entitled to large refunds on those specific tariffs.

  • If refunds go through, 2026 could become one of the most profitable years on record for many brands — while brands nationwide can also expect a new wave of spam calls from service providers eager to capture that demand.

• Marketplaces Become Media Landlords

  • 2026 forecasts show marketplaces and social commerce platforms gaining share as brands further diversify away from single-channel dependence yet remain concentrated inside a few mega-aggregators.

  • These platforms are rolling out smarter, AI-led seller/vendor scoring, catalog onboarding, and compliance tools, further tightening control over who gets visibility and at what cost.

  • Retail media will quietly morph into “agent optimization spend”: brands will buy not just ad slots but preferential treatment in AI agents (Rufus, Spark, etc.) and marketplace ranking systems, entrenching pay-to-play dynamics even deeper than today’s search ads.

  • If you don’t treat marketplaces as both distribution and paid media - with P&L guardrails for each - 2026 will be painful.

• AI-driven Personalization Widens the Gap Between Data-Rich and Data-Poor Brands

  • AI-native personalization, predictive analytics, and unified customer data are becoming baseline expectations, not differentiators, across eCommerce and retail in 2026.

  • Brands with rich first-party data and clean product information will convert better inside generative AI engines and feeds, while laggards will be optimized away by algorithms focused on conversion and margin.

  • For most brands, the most impactful “AI investment” over the next 12–18 months isn’t a cute chatbot. It’s the unglamorous work of PIM, CDP, and data governance.

• Logistics and Operations Become the Real Moat Again

  • 2026 outlooks show AI and automation driving more efficient fulfillment, tighter demand forecasting, and proactive churn prevention across leading eCommerce operations.

  • As social and agentic commerce abstract away the storefront, speed, reliability, returns, and inventory accuracy become the main levers left for brands to differentiate.

  • The loudest “innovators” next year will talk about agents and avatars; the actual winners will be the boring operators quietly fixing catalog hygiene, returns workflows, and multi‑node fulfillment while everyone else chases hype.

• Gemini Overtakes ChatGPT In Generative AI Traffic Share

  • Consider Gemini having just over 7% share 12 months ago. Fast forward to early December and that number is just under 20% share.

  • Google’s reach in Chrome, Android and streaming TV provides a captive audience to embed the offering and provide everyday seamless access to generative AI via “accidental” searches versus “destination” searches on ChatGPT.

  • Apple’s Siri won’t be replaced by Gemini but will become the secondary model for iPhone/iOS users.

• Brands Optimize for Reddit and YouTube “LLM share of voice”

  • With Reddit being the most cited source makes “LLM share of voice” a real asset for communities and brands, so expect more deliberate seeding of threads and comments aimed at being surfaced in AI answers.

  • With YouTube dominating chatbot referrals, creators and advertisers start thinking in terms of “AI-intent views” (views originating from LLM recommendations) and optimize thumbnails, titles, and chapters for AI surfaces as much as for human browse/search.

  • Current data suggests Reddit (text) and YouTube (video) remain the two default destinations representing “what real people say and show.”

If you’re leading eCommerce, retail, or marketing into 2026, the through‑line is pretty simple:

Focus less on building flashy agents and more on making your data, operations, and content so good that the agents and algorithms have no choice but to pick you.

Curious where you agree, disagree, or have your own spicy prediction — what would you add to this list for 2026?

Google's AI Comeback

The AI narrative is powerful. Earlier in 2025, Google faced serious doubts from both investors and the wider tech community about its ability to compete in generative AI. The rise of OpenAI's ChatGPT and its rapid integration into Microsoft’s ecosystem led to speculation that Google, once the undisputed AI leader, was losing its edge. This sentiment was reflected in the market: Alphabet’s share price declined over 20% following disappointing earnings and growing market share losses in both search and cloud, with headlines suggesting Google was struggling to adapt as the AI landscape shifted.

Concerns intensified as investors worried about Google ’s escalating capital expenditures on AI, which soared to record levels. High-profile product missteps, like flawed early releases of Gemini’s predecessor, only fuelled the perception that Google was falling behind.

But with the launch of Gemini 3, the story changed almost overnight. The new model dramatically outperformed earlier versions and demonstrated real progress in closing the gap with top competitors. As a result, Alphabet’s stock price surged: after unveiling Gemini 3, shares hit a record high, passing $300 for the first time ever, and recouped much of their earlier loss. For the year, the stock is now up more than 56%, as analysts and investors have renewed confidence in Google ’s AI leadership, applauding both the technology and the strategic focus behind its comeback. Gemini 3 has continued the narrative.

Key Improvements in Reasoning and Context

Gemini 3 excels in state-of-the-art reasoning and context understanding. It has topped benchmark leaderboards and outperformed previous models on complex challenges like Humanity’s Last Exam and GPQA Diamond, demonstrating near-PhD-level reasoning skills. Deep Think mode, a new enhancement, further boosts reasoning, particularly for challenging, multi-step tasks.

Multimodal and Agentic Capabilities

Unlike many earlier models, Gemini 3 is natively multimodal—it processes text, images, and code in context with each other, improving comprehension and output richness. It also introduces robust agentic features, handling complex, multi-step software tasks, autonomously validating code, and executing operations across various digital platforms. These abilities make Gemini 3 highly competitive for both consumer and developer use.

Coding and Tool Use

Gemini 3 shows dramatic gains in coding benchmarks, surpassing previous models and matching or exceeding industry leaders in real-world coding and terminal-based tasks. Its high marks on developer and terminal tool use show that Gemini 3 can independently tackle complex development jobs, making it attractive for enterprise use.

Performance versus Other Engines

• In academic reasoning, Gemini 3 has achieved substantial gains, outscoring contemporaries like GPT-5 and Claude 4.5 on the latest reasoning benchmarks.

• In multimodal understanding, Gemini 3 consistently leads, reflecting its ability to integrate and process multiple types of data.

• Coding benchmarks reveal a strong showing for Gemini 3, pushing past its forerunners and setting a new standard in code generation accuracy and reliability.

Broader Ecosystem Integration

Google has rapidly expanded Gemini 3’s presence across its product suite, including advanced generative AI modes in search, updated developer tools, and agentic workplace features. This deep integration accelerates adoption and helps close the user experience gap.

With these improvements, Gemini 3 is no longer just catching up; it now challenges the top tier of generative AI models in almost every critical area for business, academia, and creative professionals.

The New Search Reality: Why Citation Mapping Matters More Than SEO

There's a new "first page" of the internet, and it's not search results—it's AI answers. Profound mapped 4 billion AI citations across ChatGPT, Google AI Overviews, Perplexity, and other generative engines. The findings reveal what every major brand needs to understand: Where AI gets its citations determines who owns visibility in the zero-click era.

Reddit dominates as the #1 most-cited source in AI-generated answers, but the broader insight is what matters. Generative engines are building a new authority hierarchy—and it's not based on traditional SEO signals. YouTube (#1 globally), Wikipedia (#2 in the U.S.), and Reddit (#2 U.S., #3 globally) form the citation backbone, but within every category, AI selects 3-5 specific sources as authoritative.

What drives citations across generative engines?

1. Human, not corporate. AI answers lean on real talk from real people. Conversational, expert-led content consistently outpaces polished marketing. Brands trying to force viral UGC are missing the forest for the trees. Find your subject matter pockets and engage directly.

2. Clarity over popularity. Karma, upvotes, and traffic volume don't matter to LLMs. It's directness and helping people make decisions that get surfaced. The long tail wins.

3. Evergreen authority. The average AI-cited post is over a year old. Stop chasing newsjacking—start thinking about durable answers to persistent consumer questions across Reddit, YouTube, forums, and niche sites.

4. Balanced sentiment gets elevated. AI surfaces both positive and negative perspectives. Brand transparency is no longer optional—it's algorithmically required.

5. "Zero-click" is the default. Consumers are getting answers without ever visiting a site. If you're not in the trusted, cited sources, your traffic will evaporate.

Generative engines are rewriting discovery. Brands that map their citation sources, engage authentically in high-authority communities, and build durable expertise will own AI-driven visibility for years. Those that don't will be erased from consumer consideration before the purchase journey even begins.

The window to establish citation authority is open now. It won't last forever. The future of search isn't about ranking—it's about being the answer.

US Digital Advertising Market Alive and Well

AI is rapidly expanding the influence of tech giants like Google, Meta, Amazon, and others in the digital advertising market by fueling more targeted, efficient, and automated ad campaigns. As these companies double down on AI investments, their access to vast first-party data enables hyper-precise ad targeting, real-time optimizations, and stronger returns for advertisers—anchoring their dominance within the ad ecosystem.

Rapid Revenue Growth and Market Share Gains

- AI-powered advertising is helping leading platforms boost revenue by delivering ads to much more precisely defined audiences. Brands are spending more on these AI-driven platforms—such as Facebook, Instagram, YouTube, and Amazon—because they get better results.

- Today, a small group of major tech companies control nearly two-thirds of the U.S. ad market, up sharply from a decade ago, thanks mainly to AI advancements and enhanced targeting capabilities.

How AI Enhances Platform Power

- These platforms deploy AI to improve targeting, automate creative testing, optimize bids in real time, and segment audiences, driving stronger campaign performance than traditional methods.

- With programmatic, AI-driven ad buying, advertisers simply input campaign goals, budgets, and target audiences, and advanced algorithms manage the rest—from placement and bidding to creative testing—often providing greater efficiency but less transparency for marketers.

- Media companies are also leveraging AI to create more valuable first-party data segments, making their inventory increasingly attractive to brands looking for both compliance and performance.

Implications for Marketers and the Industry

- Marketers benefit from increased efficiency, higher engagement, and stronger ROI, but often have less control and insight into how their campaigns are managed as platforms operate more like black boxes.

- A handful of tech platforms now determine much of the ecosystem's rules, including how data is used, how campaigns are measured, and how ads are priced. Advertisers and smaller media players face higher risks from this level of concentration and may find it difficult to compete for budget allocations.

- This consolidation is drawing regulator attention, with growing calls for transparency and fairness in how AI-driven ad systems operate and use consumer data.

Robinhood Q3 Results: Strong Net Deposits, Gold Stickiness and ARPU Growth

Disclaimer: Long position held many times prior and actively holding now.

Robinhood announced another quarter of earnings this evening. Prior overviews are here, here, here, and here.

Assets Under Custody/Net Deposits - Assets saw an increase of 119%, an acceleration from 99% (vs 70% in the prior quarter) in the latest quarter versus last year. Keep in mind that a significant portion of the increase was due to TradePMR acquisition and the higher valuations of equity of cryptocurrency.

Gold Stickiness - The stickiness continues. Last 4 quarters adoption rate has grown from 10.5% to 12.4% to 13.1% to 14%. The subscriber base continues to grow and that stickiness leads to higher activity and lower churn. Last quarter the credit card count was over 300k but there was no reference this quarter yet the waitlist is over 3 million. Additionally, Assets Under Custody (AUC) was up 2.5x versus prior Q3.

Net Deposits - Last quarter saw a deceleration but this quarter saw a return to growth with over 29% in annual growth from prior quarter of 25%.

Average Revenue Per User (ARPU) Growth - Nearing $200 at $191 with strong year over year growth.

Bottom Line: The positive narrative remains intact and arguably accelerating. Gold subscribers continue to drive stickiness all as the revenue diversifies well outside of just crypto via credit cards, options, staking and eventual launch of banking. Robinhood continues to drive industry change and build complementary businesses within the core US market as well EU market. No reason not to remain long in my opinion despite the aggressive share price increases over the last several months years.

AWS & Advertising Power Profit Growth Despite Rising Retail Inventories

The latest quarter from Amazon shows the eCommerce giant firing on all cylinders, recording one of its most profitable periods in years. Operating margins hit double digits, continuing a steady upward trend-well ahead of last year's Q3 margin of 9.5%. Here are the key highlights:

AWS Accelerates Above Expectations
Amazon Web Services (AWS) grew 20% year-over-year, outpacing consensus and adding $5.5 billion in incremental sales. AWS is capitalizing on the industry-wide shift toward cloud, with Google and Microsoft also reporting healthy gains, albeit from smaller bases.

Marketplace and Seller Services Fuel Growth
Third-Party Seller Services delivered a second consecutive quarter of double-digit growth. Since Amazon takes a referral fee from these transactions, strong 3P sales signal not just higher volume but greater pricing power.

Advertising Revenue Momentum Continues
Amazon's advertising business jumped 24% this quarter, building on last quarter's 23% growth. This surge comes as Amazon steadily grabs market share from rivals like The Trade Desk, strengthening its position as a key force in digital marketing.

Online Sales and Inventory Imbalance
Core online sales rose 10%, the second straight quarter of double-digit growth. However, inventory levels climbed even faster: up 14% this quarter and 19% in Q2. While the pace of inventory accumulation is slowing, the overhang remains. This imbalance points to elevated pricing and stepped-up vendor funding as Amazon seeks to convert heavy inventory into margin gains.

Amazon's latest results underscore its ability to grow both the top and bottom line-even as the retail landscape remains fiercely competitive and inventory levels remain in focus.

Direct Indexing: The Smarter, Tax-Savvy Way to Invest Beyond ETFs

This comes off as a sponsored post but I can assure you it isn’t.

Over the last year, I have shifted a majority of my index investing in ETFs to direct indexing using Frec (referral) and Fidelity FidFolios. As you will in below, the case is quite compelling.

What is Direct Indexing?

Direct indexing is an investment strategy where you own the individual stocks that make up a market index—like the S&P 500 or Nasdaq—rather than investing in a pooled fund such as an ETF or mutual fund. With direct indexing, your portfolio is constructed to mirror the performance of the index by holding each underlying stock in the same weights. Thanks to advances in fractional share trading and automated platforms, direct indexing is now accessible to everyday investors, not just institutions or the ultra-wealthy.

For years, ETF investing set the standard for broad-market diversification at low cost. But direct index investing has changed the game. Instead of buying a pooled ETF, direct indexing means owning the actual stocks that comprise an index, unlocking major benefits in tax optimization and personalized control.

Why Direct Index Investing?

For years, ETF investing set the standard for anyone chasing broad-market diversification with low costs. But direct index investing has changed the conversation. Instead of buying a pooled ETF, direct indexing means owning the actual stocks that comprise an index, in the same proportions. This seemingly subtle shift unlocks major benefits, especially around tax efficiency and portfolio personalization.

The primary reason I believe so strongly in direct index investing is tax-loss harvesting.

• When markets get choppy, some stocks within your portfolio will dip—even if the overall index rises.

• Instead of selling an entire ETF and missing out on long-term winners, direct indexing lets you “harvest” losses from individual stocks, maximizing tax savings all year.

• These harvested losses can offset capital gains elsewhere, or even reduce up to $3,000 in ordinary income each year. Unused losses carry forward, offering continued tax savings into the future.

• Over time, this process—often called “tax alpha”—can boost your annual after-tax returns well beyond what’s possible with ETFs.

• Unlike ETFs, which can “trap” losses at the fund level, direct indexing gives you full access to every potential write-off.

Which brokerages provide Direct Index Investing?

There are many but I have only used Frec and Fidelity FidFolios (international). I am a big fan of the Frec product and have even sought an investment in the platform with the founder. Fidelity I use for the broader international access.

Bottom Line: Direct indexing isn’t just about replicating the market; it’s about actively optimizing for tax efficiency and keeping more of your gains.

Amazon's No Password Sharing Netflix Move

There were about 197 million people in the U.S. using Prime as of March, including some who aren’t paying for it, according to estimates from Consumer Intelligence Research Partners. Other estimates suggest there are between 140 million and 160 million U.S. Prime members. Amazon hasn’t provided a figure.

Josh Lowitz, a partner at CIRP, said his firm calculates the number of users who enjoy Prime free shipping but aren’t in the member household to be somewhere in the “low tens of millions.”

When user growth stalls, don’t allow the sharing of passwords. Netflix did it, now Amazon is running their version.

Perplexity & PayPal

PayPal and Venmo have teamed up with Perplexity, the AI company backed by NVIDIA and Jeff Bezos, to offer their users a new perk: a free one-year subscription to Perplexity Pro and early access to the Comet AI browser. Normally priced at $200 per year, this subscription gives users access to advanced AI-powered browsing features.

Absolutely brilliant way to increase monthly active users and queries whilst adding payment details for subscriptions BUT also shopping.

AI Chatbots = High Trust

Roughly 2,000 respondents from US and Germany in a recent survey revealed how search engines, AI bots, Reddit and YouTube are used and trusted. A couple of takeaways:

1. AI nearly has the trust factor of an Amazon Search

2. Search Engines still dominate with YouTube, AI chatbots, Wikipedia and Reddit now owning a sizable chunk of search:

3. The shift from Search Engines to AI is even more apparent with Gen Z:

Trade Desk: Amazon or Streaming?

Trade Desk’s stock plummeted nearly 38% following its most recent earnings release. Much of the conversation on Wall Street centers on the view that Amazon is increasingly taking market share from Trade Desk, thanks to the fast growth of Amazon's DSP platform, the rise of Prime TV, and Amazon's unique ability to easily attract small and medium-sized brands. Unlike Trade Desk, which traditionally targets larger agency groups and major brands, Amazon's advertising solution appeals to a broader client base, making it remarkably easy for smaller advertisers to launch campaigns with just a few clicks.

However, it's important to look beyond Amazon’s competitive strength. The more significant underlying challenge for Trade Desk may be the sheer volume of digital ad inventory now flooding the market. Recent developments, such as Netflix entering the ad-supported streaming space and the accelerating shift toward connected and streaming TV, have resulted in nearly a 40% increase in available ad inventory. As a result, this glut in supply has driven down the cost per mille (CPM)—the price paid to reach 1,000 viewers. Where major streaming platforms once saw CPMs above $50 in 2023, today the average falls below $30.

While competitive pressures from Amazon are real, investors considering Trade Desk should recognize that the broader market dynamics—namely, the rapid expansion of streaming ad inventory and the resulting decline in CPMs—pose a more fundamental challenge to revenue growth and profitability. I wouldn’t be a buyer at these levels.

Disclosure: No position in Trade Desk.

Inside Amazon’s Latest Quarter: Retail Rebounds, Inventory Swells, AWS Growth Slows

The latest quarter from the toll booth of eCommerce is in the books. Similar to the prior quarter, profitability remains on an upward trajectory. Amazon did not “buy” this quarter with markdowns keeping inventories at elevated levels. A few takeaways:

1. Online sales exceeded expectations, rising 11%—the fastest growth rate in years. However, inventory levels increased 19% this quarter and 15% last quarter, outpacing sales growth. This imbalance suggests elevated pricing and increased vendor funding, particularly as Amazon posted its highest operating margin in years.

2. AWS grew 17%, falling short of consensus estimates. Once a consistent outperformer, AWS is now raising concerns among Wall Street analysts.

3. Third-Party Seller Services returned to double-digit growth. Since Amazon collects a referral fee from sellers, this also points to higher pricing dynamics. Advertising revenue surged 23%—its fastest growth since Q1 2024—adding to the momentum.

   Bottom Line: While AWS may disappoint investors, the broader narrative is that Amazon is clearly leveraging price increases and vendor support to drive profitability—even as it manages elevated inventory levels.