Scale Advantage Research

169 posts

Scale Advantage Research banner
Scale Advantage Research

Scale Advantage Research

@scaleadvantage

Partner at multi-$B asset manager. All views my own and not investment advice.

New York انضم Mayıs 2026
127 يتبع61 المتابعون
تغريدة مثبتة
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
Scale creates unfair advantages. After more than a decade as a public market tech investor, this has been one of my biggest takeaways. In tech, scale can show up in many forms: e-comm logistics assets, ads signals, datacenter infrastructure, distribution defaults, R&D engines, software bundles, go-to-market reach, partner ecosystems, and highly engaged user bases. These advantages often create deep moats, strong secular growth, and superior economics. Scale is not always enough - it can breed complacency, and management execution remains critical. But more often than not, scale begets scale. That flywheel is what allows great businesses to compound for years. That's the idea behind this account: Scale Advantage. The goal is to share observations, frameworks, and case studies that are useful for investors, operators, and anyone interested in how exceptional businesses build and sustain advantage. I’ve been on Twitter/X for years as a reader. I’ve learned a lot here, and now I want to participate more actively. Feel free to DM, excited to meet + learn from others in this community. Everything I write reflects my own personal views and is not investment advice.
English
0
0
3
221
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
4/ Agents going to drive significant demand for broader non-AI compute stack. That's a bullish driver for the CSPs ($AMZN, $MSFT, $GOOG) that host these long-tail non-AI workloads (at high margins). Jeff Dean: "1 thing that these agents are going to stress is that all of our tools are too slow… often they involve interactions with tools that are designed for human latency...you want your better hardware to have low latency." Noam: "29 and a half days of those 30 days are spent waiting." Those tools (pulling data form existing software, fetching content from web servers, hitting APIs) execute on general-purpose CPUs, not accelerators. Scale to millions of agents each firing thousands of tool calls and the binding constraint shifts off the model and onto the serving + tooling stack. This is an underappreciated tailwind for general-purpose compute (CPUs, memory bandwidth, low-latency networking), not just GPUs/TPUs.
English
0
0
0
81
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
3/ Team is very bullish about algorithmic improvements. Jeff Dean: "if you look at the efficiency of, say, human learning, it's 1,000 times better than what our LLM learning can do." Koray: "The models have much more capacity than what we are getting out of them right now. There's going to be big [algorithmic] innovations that's going to enable us to do a lot more with the models."
English
1
0
1
42
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
$GOOG DeepMind's brain trust (Jeff Dean, Koray Kavukcuoglu, Noam Shazeer & Oriol Vinyals) sat down alongside the 3.5 Flash launch. They laid out several drivers for major model gains and why agents will surge non-AI workloads at AWS, Azure, and GCP. 🧵 youtube.com/watch?v=8hfpLa…
YouTube video
YouTube
English
1
0
1
902
SouthernValue
SouthernValue@SouthernValue95·
$META Alex Wang on Muse Spark and future larger models: it was a good model, we didn’t expect it to be frontier, it demonstrates our new stack is delivering “predictable scaling” in pre-training, RL, test-time, and multi-agent scaling efforts. We are cooking larger models that we are much more excited about.
English
2
2
31
5K
SouthernValue
SouthernValue@SouthernValue95·
1) $META appears to have the most incremental GW coming online of any AI lab in 2026, and including Neocloud capacity may be ahead of the ~6GW @dylan522p predicted @OpenAI and @AnthropicAI would have by EoY '26 on the @dwarkesh_sp pod (since then Ant has added @xai sublease to help close the 1GW compute gap w/ OAI) 2) MTEA potentially has less incremental inference demand competing for incremental capacity, without an exploding agent business (yet?) Based on the above, $META may have a training capacity advantage later this year, right as its new MSL team is beginning to put out good work. Is it crazy $META becomes a legitimate frontier challenger in the next 12mo?
SouthernValue tweet media
English
16
38
389
46.5K
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
$MSFT Copilot may have just hit its tipping point. Microsoft has quietly closed the gaps that made it a laggard. → Copilot Cowork - takes action across M365, now on par with Claude Cowork → Highly improved UI/design → Converged the confusing maze of products + entry points into one super app → Excel Copilot that actually delivers → Differentiated, context-aware “WorkIQ” graph underpinning All for a very compelling $30/user/month. The set-up for accelerating adoption is here.
🚨 AI News | TestingCatalog@testingcatalog

BUILD 🔥: Exclusive, an early look at Copilot Code and Copilot Cowork tabs of the upcoming super app from Microsoft, following earlier leaked screenshot of the Scout 24/7 Agent. Developers, Developers, Developers 👏

English
0
0
0
326
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
Sell-side out with their estimates of incremental profits from $META's subscription products. DB: 4-13% EPS upside Wolfe: 8% EPS upside Wells Fargo: 12% EPS upside This is the biggest new rev/EBIT stream announced YTD.
English
0
0
2
171
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
Thank you. On long-term penetration rates, certainly possible as they add more features to the bundle as Noami Gleit alluded to in announcement. I do worry META's higher mix of emerging market users (more than half of user base) and older demo vs. SNAP caps some of the upside. But I may be underestimating the mix of quasi-influencers on IG (vs. SNAP) that would find this very valuable.
English
0
0
0
106
Five Points Capital
Five Points Capital@fivepointscap·
Very nice work! I think penetration rate longer term on Instagram Plus could easily be 10%+. And then Meta One for Business or whatever it’s called exactly is going to be massive. I just can’t stress enough how valuable that subscription will be to SMBs. Digital ads, brand management, content creation, customer service (via WhatsApp), and then AI assistant/agentic tools. How could you not have that?
English
1
0
3
325
Five Points Capital
Five Points Capital@fivepointscap·
Some thoughts on the $META news today. This starts a new era for Meta. They’re no longer just a social media advertising company. Meta One for business will be a goldmine. For $20/mo, you’ll get IG Plus, FB Plus, WA Plus, and the enhanced LLM. If you’re a small business, this is such a no-brainer vs. $20/mo for GPT or Claude. IG and FB for advertising and brand management, WA for customer communication (handled by AI), and then the LLM to help you with everything else for your business. Maybe they’ll launch a $250/mo tier… still peanuts for a decent small business. You then add in ALL of the young people who want to be an influencer, content creator, or just build their own brand in some way. Anyone who actually wants to pursue that will be paying the $8/mo at least and that’s millions and millions of people. Then on the consumer side… I’m sure $META will be able to integrate the LLM into Instagram/Facebook/WhatsApp in such a way that they get billions of users using their AI tools. And like I’ve said before, the consumer AI game is about engagement, which Meta are the masters of. I also think Instagram Plus will instantly become a status symbol the moment it’s launched, and will have 100M+ users quickly. I’m not even going to begin to try to forecast the revenue and earnings this will generate until we have more details. But this is a total game changer for $META and it answers the capex questions beautifully.
amit@amitisinvesting

$META META IS GETTING INTO THE AI SUBSCRIPTION GAME. - Meta is expanding paid subscriptions globally across Instagram, Facebook, and WhatsApp, while also testing new AI, creator, and business plans under the “Meta One” brand. - Consumer plans are expected to start around $2.99 to $3.99 per month. - Meta is also testing AI subscription tiers at $7.99 and $19.99 per month, with the higher tier offering more capacity for complex prompts, deeper reasoning, and increased image and video generation. Looks like Zucks is deciding that his next core revenue stream will be subscriptions beyond advertising... $META +3%

English
15
4
153
26.8K
Couch Investor🛋️
Couch Investor🛋️@Couch_Investor·
$SNAP has over 25 million paying Snapchat+ subscribers. That's ~2.6% of MAUs and generates over $1 billion in annual revenue. If $META manages to convert ~2.6% of their 3.6 billion MAUs, that would result in an extra $4.48 billion in high-margin revenue/year. That's a 2% to 3% revenue increase.
Couch Investor🛋️ tweet media
English
6
2
89
8.1K
Benny Blanc◎
Benny Blanc◎@beniaminoblanco·
@scaleadvantage @evrgn11112231 @SupBagholder The fact that $META multiple has been punished by investors for aggressive/"risky" capex is largely a reflection that the market thinking re: the impact of AI has not caught up yet. Zuck was several moves ahead. Now shareholders just need to wait for the market to catch up. ⏱️
English
1
0
0
58
Supreme Bagholder
Supreme Bagholder@SupBagholder·
What the market doesn't understand is that the Anthropic x SpaceX deal adds a downside protection for $META. SpaceX rents out 300+ MW (220k GPUs) of compute for 15B/year. $META owns >1.3M GPUs as of 2025 YE.
English
14
8
229
30.7K
Rihard Jarc
Rihard Jarc@RihardJarc·
Did people really expect $META wouldn't charge a subscription for AI products and use their +3B distribution network? $SNAP makes $5B in ad revenue and $1B in direct subs revenue (growing 87% YoY). $META can add a few $10B to its bottom line just by providing solid-value subscriptions. Distribution is the ultimate AI bottleneck and moat, as we have AI models and services available in every app and platform.
English
27
32
466
73.1K
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
Couple pushbacks: - Meta Verified goes after creators vs these features have more mass market appeal and the price point limited adoption at $12-15/mo vs this is $4/mo. - I don't think you're right that Instagram Plus is mostly existing features behind a paywall, see below list. - The added features seem pretty compelling. NEW functionality: 1/ Aggregate story rewatch counts 2/ Multiple custom audience lists (only Close Friends exists today) 3/ Story spotlight boost 4/ Extending stories past 24 hours 5/ Incognito story viewing 6/ Posting to profile without hitting follower feeds 7/ Super Heart reactions 8/ Custom app icons 9/ Customizable bio fonts EXISTING, just expanded: 10/ More profile pins (currently capped at 3) AMBIGUOUS: 11/ Story viewer list search That's 9 of 11 net new, 1 expansion, 1 unclear.
English
0
0
3
246
Eric Seufert
Eric Seufert@eric_seufert·
$SNAP’s subscription product is not a useful benchmark for $META’s in estimating potential subscription revenue. First, Meta already has a subscription product: Meta Verified. That’s the most direct opportunity for establishing a conversion baseline. It’s low. One might also look to Meta’s subscription product in the EU, required through the DMA (<1% adoption). But more fundamentally: Snap takes an entirely different approach with subscriptions than what Meta is proposing. Snap’s ad platform lags those of its peers in terms of monetization efficiency. On an ARPU basis, Snap is incentivized to nudge users into the subscription by shifting core functionality behind the subscription payment. That is profitable, even if some users churn completely when faced with paying or not having access to core functionality. That approach would likely not be profitable for Meta, given its ARPU / subscription price differential. So Meta can’t gate core functionality behind the subscription; it must find new functionality to offer to entice subscriptions. How effective will that be? Almost certainly less effective than Snap’s approach at driving subscriptions, given the diminished value proposition. So assume a lower subscription rate. Currently Meta’s subscription functionality seems aimed at influencers (Reels view metrics, new publishing modes), which represent a tiny proportion of the audience. So simply applying Snap subscription rates to the Meta product will likely substantially overstate subscription revenue. Important also to note the scale of Meta’s high-margin ads business: likely ~$240BN in 2026 revenues.
English
2
1
53
7.3K
Scale Advantage Research
Scale Advantage Research@scaleadvantage·
I don't think you're right that Instagram Plus is mostly existing features behind a paywall. Going through the list: NEW functionality: 1/ Aggregate story rewatch counts 2/ Multiple custom audience lists (only Close Friends exists today) 3/ Story spotlight boost 4/ Extending stories past 24 hours 5/ Incognito story viewing 6/ Posting to profile without hitting follower feeds 7/ Super Heart reactions 8/ Custom app icons 9/ Customizable bio fonts EXISTING, just expanded: 10/ More profile pins (currently capped at 3) AMBIGUOUS: 11/ Story viewer list search That's 9 of 11 net new, 1 expansion, 1 unclear. This is not like SNAP where they moved existing functionality behind a paywall. techcrunch.com/2026/05/27/met…
English
0
1
1
178
Eric Seufert
Eric Seufert@eric_seufert·
Meta announced Instagram / Facebook Plus in January has been testing them since March. If these products are moving into GA, the conversion rate must be at least somewhat promising. Beyond the revenue opportunity, which isn’t material if you imagine the company does ~$240BN in ad revenue in 2026, I imagine Meta is feeling pressure to find an obvious example of monetizing consumer AI use. I think they could find a better and more compelling argument from the advertising angle if they could improve their storytelling capacity there. But Zuckerberg seems unwilling to champion that, so yet another subscription is offered (Meta Verified is the existing subscription that, curiously, remains separate from this).
Rihard Jarc@RihardJarc

Did people really expect $META wouldn't charge a subscription for AI products and use their +3B distribution network? $SNAP makes $5B in ad revenue and $1B in direct subs revenue (growing 87% YoY). $META can add a few $10B to its bottom line just by providing solid-value subscriptions. Distribution is the ultimate AI bottleneck and moat, as we have AI models and services available in every app and platform.

English
3
0
53
28.2K
Shay Boloor
Shay Boloor@StockSavvyShay·
$META is reportedly testing paid subscription tiers across its apps: • Instagram Plus: $3.99/month • Facebook Plus: $3.99/month • WhatsApp Plus: $2.99/month Meta is turning extra app features into another recurring revenue layer across its massive user base.
Shay Boloor tweet mediaShay Boloor tweet media
English
348
246
2.1K
1.2M