Samsung just became the first tech company to clear ₩100tn of operating profit in a quarter: ₩107.4tn (over $75B), with chips making ₩106.5tn of it.
The stock fell 4%. Buy calls made the night before are underwater.
Every call, graded:
buzzberg.ai/ticker/005930.…
Seoul Desk scorecard
Calls we logged in the week before results (Oct 22-28): 66 buy, 104 watch, 2 stay away, 0 bet against
How the named calls look at the Oct 29 close (₩232,000):
• Lee Ji-hwan, buy, Oct 22: +8.2%
• Park Byeong-chang, buy as a defensive leader, Oct 26: +8.9%
• Lee Kwon-hee, buy on HBM4, Oct 28: -4.1%
• Lee Ju-yeon, buy ahead of shareholder returns, Oct 28: -4.1%
• Ha Chang-wan, stay away, Oct 23: +7.7%
The numbers matched what Samsung flagged on Oct 8, and HBM4 sales more than tripled from Q2.
Every buy call from Oct 22 to 27 is up 5% to 9%. Every one from Oct 28 is down 4%.
Risk: net profit came in at ₩75.1tn vs an LSEG estimate of about ₩84tn, a miss of over ₩8tn, and the stock ran about 15% from the Oct 15 low into the print.
Buying this dip or waiting? We check every call again on Nov 12, two weeks after the print.
Follow to see who was right!
The stock fell 4%. Buy calls made the night before are underwater.
Every call, graded:
buzzberg.ai/ticker/005930.…
Seoul Desk scorecard
Calls we logged in the week before results (Oct 22-28): 66 buy, 104 watch, 2 stay away, 0 bet against
How the named calls look at the Oct 29 close (₩232,000):
• Lee Ji-hwan, buy, Oct 22: +8.2%
• Park Byeong-chang, buy as a defensive leader, Oct 26: +8.9%
• Lee Kwon-hee, buy on HBM4, Oct 28: -4.1%
• Lee Ju-yeon, buy ahead of shareholder returns, Oct 28: -4.1%
• Ha Chang-wan, stay away, Oct 23: +7.7%
The numbers matched what Samsung flagged on Oct 8, and HBM4 sales more than tripled from Q2.
Every buy call from Oct 22 to 27 is up 5% to 9%. Every one from Oct 28 is down 4%.
Risk: net profit came in at ₩75.1tn vs an LSEG estimate of about ₩84tn, a miss of over ₩8tn, and the stock ran about 15% from the Oct 15 low into the print.
Buying this dip or waiting? We check every call again on Nov 12, two weeks after the print.
Follow to see who was right!
0
0
1
236
Caterpillar $CAT is down 17% since Oct 6 and reports Thursday. A photonics researcher we track holds a starter position anyway, betting AI reaches it two ways: self-driving machines and data-center power.
Their call, and every $CAT call we track:
buzzberg.ai/ticker/CAT?utm…
First Call
The case, from Gaetano's (@gaetano2026) Oct 24 Substack:
• Autonomous machines doing useful work in the real world
• A dealer network: each machine sold can lead to years of parts and service work
• Power generation: engines and turbines going into data centers
The power leg already shows in the numbers: Caterpillar's Power & Energy sales rose 17% to $8.24B in Q2.
It's Gaetano's first buy call on $CAT in our records. They rank #72 on our 30-day board, which grades their calls, not anyone's returns.
Thursday's test is Power & Energy sales. Growth near Q2's 17% keeps the power leg intact; a sharp slowdown weakens it. Autonomy is a longer story that one quarter won't settle.
We'll put Thursday's number next to their case. Follow to see whether it held!
Their call, and every $CAT call we track:
buzzberg.ai/ticker/CAT?utm…
First Call
The case, from Gaetano's (@gaetano2026) Oct 24 Substack:
• Autonomous machines doing useful work in the real world
• A dealer network: each machine sold can lead to years of parts and service work
• Power generation: engines and turbines going into data centers
The power leg already shows in the numbers: Caterpillar's Power & Energy sales rose 17% to $8.24B in Q2.
It's Gaetano's first buy call on $CAT in our records. They rank #72 on our 30-day board, which grades their calls, not anyone's returns.
Thursday's test is Power & Energy sales. Growth near Q2's 17% keeps the power leg intact; a sharp slowdown weakens it. Autonomy is a longer story that one quarter won't settle.
We'll put Thursday's number next to their case. Follow to see whether it held!
0
0
3
250
We asked on Monday (post below) whether whether Q3 at Samsung Electro-Mechanics (SEMCO) would talk about price or only volume. It did, in a record quarter: operating profit up 165% on a year ago.
Yet 11 of the 14 investors who said buy (Oct 1-24) are still underwater.
Revenue rose 30%, substrate sales 58%. Management says it's in price talks on chip substrates, citing tight capacity and costlier materials, and guided higher prices for Q4 and 2027.
Why most are red: they first said buy in early October, before a slide of over 30%. Even after today's jump, they're down 16% to 22%.
The 3 in the green came Oct 16 or later, including Chesley Investment Advisory's Oct 24 panel: up 17% since we logged it.
Monday's stay-away voice, Lee Ji-hwan, called it expensive Oct 5; it's still 22% lower. Their Oct 24 call: rotate from power and MLCC names into semis. Up 17% since.
Risk: new capacity adds meaningful revenue only from 2028, and price talks aren't signed hikes.
Round one to the bull case. Round two is the Q4 price talks. Bulls or the stay-away: who takes it?
Seoul Desk: Korea's market calls, logged before results and scored after. Samsung Electronics reports Thursday. Follow for its scorecard.
x.com/buzzberg_ai/st…
Yet 11 of the 14 investors who said buy (Oct 1-24) are still underwater.
Revenue rose 30%, substrate sales 58%. Management says it's in price talks on chip substrates, citing tight capacity and costlier materials, and guided higher prices for Q4 and 2027.
Why most are red: they first said buy in early October, before a slide of over 30%. Even after today's jump, they're down 16% to 22%.
The 3 in the green came Oct 16 or later, including Chesley Investment Advisory's Oct 24 panel: up 17% since we logged it.
Monday's stay-away voice, Lee Ji-hwan, called it expensive Oct 5; it's still 22% lower. Their Oct 24 call: rotate from power and MLCC names into semis. Up 17% since.
Risk: new capacity adds meaningful revenue only from 2028, and price talks aren't signed hikes.
Round one to the bull case. Round two is the Q4 price talks. Bulls or the stay-away: who takes it?
Seoul Desk: Korea's market calls, logged before results and scored after. Samsung Electronics reports Thursday. Follow for its scorecard.
x.com/buzzberg_ai/st…
Samsung Electro-Mechanics is spending ₩6.78T on chip substrates. Customers guaranteed the volume. The plan says nothing about price.
Every call we logged, and why 14 investors we track still say buy:
buzzberg.ai/ticker/009150.…
Seoul Desk: what Korean investors are saying
The plan: two new substrate factories (Sep 29 release)
- Sejong, Korea: ₩4.27T, its biggest single-product investment ever. Mass production starts Sep 2028
- Vietnam: about ₩2.51T, running by June 2030
- Customer funding and volume guarantees. No customer named. Nothing on price
So volume is promised, and price is the open question. It isn't the only thing that moves the stock, but it's the one this post looks at. Pricing helps determine what that promised volume earns.
The bull case: prices are already going up.
A Chesley Investment Advisory panel (Oct 22):
- Supply is so tight that makers are raising prices even on existing products
- This year's second-half increases come from scarcity, not from passing on costs
- Order backlogs jumped from the prior quarter
Even they said be picky: new 2027-28 capacity could squeeze margins if demand slows.
The cautious case: it has already run a long way.
Lee Ji-hwan, Aurora Investment Advisory (Oct 24), isn't betting against it. After its run, the stock trades at far higher multiples than Samsung Electronics, so in a pullback they would trim it and buy the chipmakers.
Investors we track on the stock, Oct 1-24: 14 buy, 0 bet against, 2 stay away.
How to read what comes next:
- Substrate prices still rising: support for the bull case, since the same volume earns more
- Prices flattening while the new lines are still being built: support for the cautious case
The first tell is the company's third-quarter results: do they talk about price, or only volume?
The bigger one comes later, in the fourth-quarter substrate price talks.
0
0
1
269
Half of the top 10 channels on our board are crypto shows. The best-ranked one has 15.6K followers and sits above a show with 649K.
We rank channels on what their calls returned, not on audience size. Check how your favourite YouTube channel performs:
buzzberg.ai/speakers?board… he top 10 (Oct 27 snapshot, average return per call within 30 days):
• #4 Delphi Digital (@Delphi_Digital), 15.6K followers: +5.4% across 60 graded calls
• #5 Thread Guy (@notthreadguy): +3.1% across 439
• #8 Unchained's Chopping Block (@Unchained_pod): +2.6% across 312
• #9 Anthony Pompliano (@APompliano), 649K followers: +3.1% across 350
• #10 Empire (@theempirepod): +2.7% across 151
Returns are averages per graded call, not anyone's trading returns.
Which channel do you watch, and where do you think it ranks?
Follow @buzzberg_ai for the next board.
We rank channels on what their calls returned, not on audience size. Check how your favourite YouTube channel performs:
buzzberg.ai/speakers?board… he top 10 (Oct 27 snapshot, average return per call within 30 days):
• #4 Delphi Digital (@Delphi_Digital), 15.6K followers: +5.4% across 60 graded calls
• #5 Thread Guy (@notthreadguy): +3.1% across 439
• #8 Unchained's Chopping Block (@Unchained_pod): +2.6% across 312
• #9 Anthony Pompliano (@APompliano), 649K followers: +3.1% across 350
• #10 Empire (@theempirepod): +2.7% across 151
Returns are averages per graded call, not anyone's trading returns.
Which channel do you watch, and where do you think it ranks?
Follow @buzzberg_ai for the next board.
0
0
1
262
SK Hynix kept ₩80 of every ₩100 in sales as operating profit last quarter.
It sells memory chips, which is a textbook commodity.
A year ago that margin was 47%. Operating profit is up 632%, to ₩83.3T in three months.
Going in, 45 of the 79 speakers we track refused to buy. oth of their tests passed overnight, check who moves first 👇
buzzberg.ai/ticker/000660.…
It sells memory chips, which is a textbook commodity.
A year ago that margin was 47%. Operating profit is up 632%, to ₩83.3T in three months.
Going in, 45 of the 79 speakers we track refused to buy. oth of their tests passed overnight, check who moves first 👇
buzzberg.ai/ticker/000660.…
45 of the 79 speakers we track on SK Hynix went into its results with no buy or sell call. Two said exactly what would change that: proof HBM margins hold, and a real buyback plan.
The report answered both.
Who was already buying going in, and why:
buzzberg.ai/ticker/000660.…
The count, Oct 20-27 (latest call per speaker, Seoul-listed shares): 28 buy, 45 watch, 6 stay away, 0 bet against. Mostly Korean market shows on YouTube.
Going in: on Oct 23, SK Hynix dropped its US listing plan for Solidigm, its storage unit. Seoul shares closed up 4.4%; several shows called it one dilution worry gone.
The two tests:
• Yoon Jin (815 Money Talk, Oct 24): HBM4 could squeeze margins; wants proof they hold before calling a rebound
• Park Myung-seok (3PRO TV, Oct 23): a buyback mostly puts a floor under the stock; wants a concrete shareholder-return plan
The report:
• Margins: 80% operating margin, vs 76% in Q2, as HBM4 shipments ramped up
• Shareholder returns: about 100 trillion won planned for 2026
Next: whether these two, and the other 43 on the fence, turn buyers.
Follow @buzzberg_ai to see who moves first 👀
0
0
1
247
45 of the 79 speakers we track on SK Hynix went into its results with no buy or sell call. Two said exactly what would change that: proof HBM margins hold, and a real buyback plan.
The report answered both.
Who was already buying going in, and why:
buzzberg.ai/ticker/000660.…
The count, Oct 20-27 (latest call per speaker, Seoul-listed shares): 28 buy, 45 watch, 6 stay away, 0 bet against. Mostly Korean market shows on YouTube.
Going in: on Oct 23, SK Hynix dropped its US listing plan for Solidigm, its storage unit. Seoul shares closed up 4.4%; several shows called it one dilution worry gone.
The two tests:
• Yoon Jin (815 Money Talk, Oct 24): HBM4 could squeeze margins; wants proof they hold before calling a rebound
• Park Myung-seok (3PRO TV, Oct 23): a buyback mostly puts a floor under the stock; wants a concrete shareholder-return plan
The report:
• Margins: 80% operating margin, vs 76% in Q2, as HBM4 shipments ramped up
• Shareholder returns: about 100 trillion won planned for 2026
Next: whether these two, and the other 43 on the fence, turn buyers.
Follow @buzzberg_ai to see who moves first 👀
The report answered both.
Who was already buying going in, and why:
buzzberg.ai/ticker/000660.…
The count, Oct 20-27 (latest call per speaker, Seoul-listed shares): 28 buy, 45 watch, 6 stay away, 0 bet against. Mostly Korean market shows on YouTube.
Going in: on Oct 23, SK Hynix dropped its US listing plan for Solidigm, its storage unit. Seoul shares closed up 4.4%; several shows called it one dilution worry gone.
The two tests:
• Yoon Jin (815 Money Talk, Oct 24): HBM4 could squeeze margins; wants proof they hold before calling a rebound
• Park Myung-seok (3PRO TV, Oct 23): a buyback mostly puts a floor under the stock; wants a concrete shareholder-return plan
The report:
• Margins: 80% operating margin, vs 76% in Q2, as HBM4 shipments ramped up
• Shareholder returns: about 100 trillion won planned for 2026
Next: whether these two, and the other 43 on the fence, turn buyers.
Follow @buzzberg_ai to see who moves first 👀
0
0
1
905
How we calculate an author's returns:
1. A return starts at the price of the post that opens the call: the author's explicit bullish call on the ticker
2. Not every bullish post becomes a call. It has to be clear: "I bought more", "this is a buy", or a firm upside view like the market isn't pricing this in". Praise, upbeat takes, news and mentions stay Watch
3. Later bullish posts on the same ticker are Long Hold and never move that entry
4. A call ends at the price of the post where the author says they've closed the position or turns bearish. If nothing ends it sooner, it closes after 360 days
5. Each call is checked 7, 30, 90, 180 and 360 days after the entry, against the S&P 500 over the same dates
Video explanation below
1. A return starts at the price of the post that opens the call: the author's explicit bullish call on the ticker
2. Not every bullish post becomes a call. It has to be clear: "I bought more", "this is a buy", or a firm upside view like the market isn't pricing this in". Praise, upbeat takes, news and mentions stay Watch
3. Later bullish posts on the same ticker are Long Hold and never move that entry
4. A call ends at the price of the post where the author says they've closed the position or turns bearish. If nothing ends it sooner, it closes after 360 days
5. Each call is checked 7, 30, 90, 180 and 360 days after the entry, against the S&P 500 over the same dates
Video explanation below
1
0
2
241
Crypto bears are getting louder. Bearish calls on $BTC and $ETH jumped from 14 (Sep 29 to Oct 5) to 22 (Oct 20 to 26). Bullish calls barely moved: 89 to 94.
Last 24 hours: 12 bearish $BTC call records vs 7 bullish.
Who's on which side, by name:
buzzberg.ai/ticker/BTC?utm…
The bitcoin. @au_xbt is short into the Fed but waiting to cover. Ansem (@blknoiz06) is bearish for now, yet still expects another leg higher later.
The bull: Galaxy's Alex Thorn (@intangiblecoins), who said Monday it's time to add and hold.
Have we seen this before? We checked four bear waves this year, weeks when over a third of $BTC calls were bearish:
• Twice, BTC then fell: 26% after Jan 25, 18% after May 20
• Twice, BTC then rose: 9% after Mar 25, 22% after Aug 3
(Moves over the next 30 days)
Same signal, opposite endings. Today bears are about 21% of the week's calls, still below that line.
The Fed decides Wednesday. Bulls or bears?
After the Fed, we grade both sides, by name. Follow @buzzberg_ai to see who called it 👀
Last 24 hours: 12 bearish $BTC call records vs 7 bullish.
Who's on which side, by name:
buzzberg.ai/ticker/BTC?utm…
The bitcoin. @au_xbt is short into the Fed but waiting to cover. Ansem (@blknoiz06) is bearish for now, yet still expects another leg higher later.
The bull: Galaxy's Alex Thorn (@intangiblecoins), who said Monday it's time to add and hold.
Have we seen this before? We checked four bear waves this year, weeks when over a third of $BTC calls were bearish:
• Twice, BTC then fell: 26% after Jan 25, 18% after May 20
• Twice, BTC then rose: 9% after Mar 25, 22% after Aug 3
(Moves over the next 30 days)
Same signal, opposite endings. Today bears are about 21% of the week's calls, still below that line.
The Fed decides Wednesday. Bulls or bears?
After the Fed, we grade both sides, by name. Follow @buzzberg_ai to see who called it 👀
0
0
10
387
Anthropic's S-1 warns AI could pose catastrophic or existential risks. It also commits Anthropic to $265.6B of compute, owed whether it uses it or not.
At Dec 31, that bill was $54.6B.
Every dated call on $ANT, bulls and stay-aways:
buzzberg.ai/ticker/ANTHROP…
The bull case is real. H1 2026 revenue grew 14x to $16.29B. Cash and short-term investments: $131.7B. About 90% of the $80.7B loss is a non-cash charge.
Across all the history we track, Anthropic drew 67 bullish ideas, 32 stay-aways and 0 bets against, from 198 speakers.
One of our best-ranked investors is in the stay-away camp. AlmaCap (#6 on our 30-day board) flagged Anthropic as an avoid on Oct 8, arguing Microsoft and Meta were weaning off it. @bubbleboi (#62, Sep 29) and @santiagoroel (#179, Oct 16) also called it a stay-away on valuation. All three calls came before the S-1.
$GOOGL and $AMZN hold about $224B of those minimums, so their payday depends on Anthropic's revenue keeping up with the bill. That's the line to watch in every filing after the listing.
We'll post the scorecard when $ANT lists.
The bulls or the best-ranked skeptic: who's right on $ANT?
At Dec 31, that bill was $54.6B.
Every dated call on $ANT, bulls and stay-aways:
buzzberg.ai/ticker/ANTHROP…
The bull case is real. H1 2026 revenue grew 14x to $16.29B. Cash and short-term investments: $131.7B. About 90% of the $80.7B loss is a non-cash charge.
Across all the history we track, Anthropic drew 67 bullish ideas, 32 stay-aways and 0 bets against, from 198 speakers.
One of our best-ranked investors is in the stay-away camp. AlmaCap (#6 on our 30-day board) flagged Anthropic as an avoid on Oct 8, arguing Microsoft and Meta were weaning off it. @bubbleboi (#62, Sep 29) and @santiagoroel (#179, Oct 16) also called it a stay-away on valuation. All three calls came before the S-1.
$GOOGL and $AMZN hold about $224B of those minimums, so their payday depends on Anthropic's revenue keeping up with the bill. That's the line to watch in every filing after the listing.
We'll post the scorecard when $ANT lists.
The bulls or the best-ranked skeptic: who's right on $ANT?
@AnthropicAI say AI poses existential risks to humanity and then spend $750B on accelerating it
2
1
8
1,278
Nebius $NBIS secured $3.5B from Citi just as Elon Musk warned that neoclouds face a financing squeeze.
The money is available. The question is what comes after this loan.
Compare the dated investor calls and their reasoning:
buzzberg.ai/ticker/NBIS?ut…
The facility, per Nebius’s filing:
* 364-day maturity
* Initial rate: Term SOFR + 2.75%
* Margin rises 0.25 points every 90 days, up to SOFR + 3.50%
* Maturity: October 16, 2027
The next test is whether Nebius can replace short-term financing with longer-term capital, and on what terms.
For context, CoreWeave’s $CRWV 9.75% notes, priced in April, yielded about 15.8% on October 15.
A traded bond and a secured bank loan are different instruments. But the bond shows how sharply investors have repriced one neocloud borrower’s debt.
SpaceX is taking a different route: a binding commitment letter for a GPU equipment lease with an Apollo-managed entity, with about $40B of expected payments. It is still subject to closing conditions.
On Buzzberg, two investors see different paths for Nebius:
* @TheValueist: an Oct 24 post, which Buzzberg logged as an avoid, pointed to financing costs and the short maturity weighing on near-term equity returns. They also acknowledged Nebius’s access to capital.
* @daniel_koss: favors Nebius over the neocloud basket, arguing that stronger companies could benefit from consolidation. On Oct 26, they said they weren’t selling and would add on further weakness.
Nebius says its investment-grade customer contracts, more than $40B of them, give it another potential source of financing. That remains a claim to watch as future deals are arranged.
$NBIS was Buzzberg’s #2 most-mentioned ticker on Oct 26, with 74 tracked mentions, behind $NVDA’s 92.
The question: can Nebius lock in longer-term financing on attractive terms ahead of the loan’s first margin step-up in January?
The money is available. The question is what comes after this loan.
Compare the dated investor calls and their reasoning:
buzzberg.ai/ticker/NBIS?ut…
The facility, per Nebius’s filing:
* 364-day maturity
* Initial rate: Term SOFR + 2.75%
* Margin rises 0.25 points every 90 days, up to SOFR + 3.50%
* Maturity: October 16, 2027
The next test is whether Nebius can replace short-term financing with longer-term capital, and on what terms.
For context, CoreWeave’s $CRWV 9.75% notes, priced in April, yielded about 15.8% on October 15.
A traded bond and a secured bank loan are different instruments. But the bond shows how sharply investors have repriced one neocloud borrower’s debt.
SpaceX is taking a different route: a binding commitment letter for a GPU equipment lease with an Apollo-managed entity, with about $40B of expected payments. It is still subject to closing conditions.
On Buzzberg, two investors see different paths for Nebius:
* @TheValueist: an Oct 24 post, which Buzzberg logged as an avoid, pointed to financing costs and the short maturity weighing on near-term equity returns. They also acknowledged Nebius’s access to capital.
* @daniel_koss: favors Nebius over the neocloud basket, arguing that stronger companies could benefit from consolidation. On Oct 26, they said they weren’t selling and would add on further weakness.
Nebius says its investment-grade customer contracts, more than $40B of them, give it another potential source of financing. That remains a claim to watch as future deals are arranged.
$NBIS was Buzzberg’s #2 most-mentioned ticker on Oct 26, with 74 tracked mentions, behind $NVDA’s 92.
The question: can Nebius lock in longer-term financing on attractive terms ahead of the loan’s first margin step-up in January?
0
2
12
923
Anthropic $ANT lost $80.7B in six months. About 90% of that was a paper charge, not cash.
The number to watch is the $265.6B in minimum purchase commitments, primarily hosting and infrastructure.
Revenue grew 14x: $16.29B in H1 2026, versus $1.16B a year earlier.
Operating ng cash flow was positive $3.9B. The paper charge is $72.4B of fair-value changes on convertible notes, derivatives and SAFEs.
Now the landlord side:
Minimum purchase commitments reached $265.6B at June 30, up from $54.6B at December 31.
Separately, the filing discloses minimum spending of approximately:
- $114.5B with Google $GOOGL through July 2033
- $110B with Amazon $AMZN through April 2036
Those agreements require usage-independent minimum payments, subject to contractual exceptions. Google’s figure includes later amendments; these aren’t a breakdown of the June total.
Our read: the suppliers gain contractual revenue visibility. Anthropic must turn the capacity into profitable customer demand.
@BobEUnlimited raised that growth-versus-buildout question on October 9.
We checked the original.
AlmaCap (@almacap114204), #6 on our 30-day-return leaderboard, raised a related concern on October 15: Anthropic could catch up on efficiency but struggle with its own distribution. They preferred $META, $GOOGL, $AMZN, and $MSFT.
This filing adds evidence, not a verdict: first-half results cannot settle their argument about growth after summer.
The countercase is strong customer expansion and improving compute economics. And contractual revenue alone doesn't make Google or Amazon attractive at any valuation.
Research dated investment theses through Buzzberg MCP:
buzzberg.ai/mcp?ui_lang=en…
Can Anthropic grow into the bill before the bill outruns growth?
The number to watch is the $265.6B in minimum purchase commitments, primarily hosting and infrastructure.
Revenue grew 14x: $16.29B in H1 2026, versus $1.16B a year earlier.
Operating ng cash flow was positive $3.9B. The paper charge is $72.4B of fair-value changes on convertible notes, derivatives and SAFEs.
Now the landlord side:
Minimum purchase commitments reached $265.6B at June 30, up from $54.6B at December 31.
Separately, the filing discloses minimum spending of approximately:
- $114.5B with Google $GOOGL through July 2033
- $110B with Amazon $AMZN through April 2036
Those agreements require usage-independent minimum payments, subject to contractual exceptions. Google’s figure includes later amendments; these aren’t a breakdown of the June total.
Our read: the suppliers gain contractual revenue visibility. Anthropic must turn the capacity into profitable customer demand.
@BobEUnlimited raised that growth-versus-buildout question on October 9.
We checked the original.
AlmaCap (@almacap114204), #6 on our 30-day-return leaderboard, raised a related concern on October 15: Anthropic could catch up on efficiency but struggle with its own distribution. They preferred $META, $GOOGL, $AMZN, and $MSFT.
This filing adds evidence, not a verdict: first-half results cannot settle their argument about growth after summer.
The countercase is strong customer expansion and improving compute economics. And contractual revenue alone doesn't make Google or Amazon attractive at any valuation.
Research dated investment theses through Buzzberg MCP:
buzzberg.ai/mcp?ui_lang=en…
Can Anthropic grow into the bill before the bill outruns growth?
1
4
7
1,463
Brazil's market spent October betting on a Bolsonaro win. Lula won, and the Brazil fund $EWZ was posted down about 20% overnight.
Every call on $EWZ, before and after the vote:
buzzberg.ai/ticker/EWZ?utm…
Lula took Sunday's runoff with 51.8% of valid votes to Flávio Bolsonaro's 48.2%. Real futures fell about 9% overnight.
Why the panic is about money, not politics: investors doubt a fourth Lula term brings the spending cuts they want, with public debt near 93% of GDP. Weaker budget discipline keeps borrowing costs higher, which weighs on stock valuations. A lower price alone doesn't fix that.
A Citi survey of 73 investors had expected a Lula win to knock the real down at least 5%. The futures moved almost twice that.
What the investors we track said:
- Oct 21-25, before the result: 2 buy, 0 bet against, 0 stay away
- Oct 25 23:00 UTC on, as results came in: 0 buy, 1 bet against, 3 stay away, 11 watching without a call
Our read: nobody we track is buying the drop yet. The useful counterpoint: @Mr_Derivatives notes the fall only takes $EWZ back to where it traded two months ago, and suggests giving it days or weeks to settle. That's perspective, not a buy call.
What would change the case: a cheaper price, a clearer fiscal plan, or both. Brazil's market reopens today, that gives the first real read.
We pulled every call above with one question to the Buzzberg MCP inside Claude: buzzberg.ai/mcp?utm_source…
Every call on $EWZ, before and after the vote:
buzzberg.ai/ticker/EWZ?utm…
Lula took Sunday's runoff with 51.8% of valid votes to Flávio Bolsonaro's 48.2%. Real futures fell about 9% overnight.
Why the panic is about money, not politics: investors doubt a fourth Lula term brings the spending cuts they want, with public debt near 93% of GDP. Weaker budget discipline keeps borrowing costs higher, which weighs on stock valuations. A lower price alone doesn't fix that.
A Citi survey of 73 investors had expected a Lula win to knock the real down at least 5%. The futures moved almost twice that.
What the investors we track said:
- Oct 21-25, before the result: 2 buy, 0 bet against, 0 stay away
- Oct 25 23:00 UTC on, as results came in: 0 buy, 1 bet against, 3 stay away, 11 watching without a call
Our read: nobody we track is buying the drop yet. The useful counterpoint: @Mr_Derivatives notes the fall only takes $EWZ back to where it traded two months ago, and suggests giving it days or weeks to settle. That's perspective, not a buy call.
What would change the case: a cheaper price, a clearer fiscal plan, or both. Brazil's market reopens today, that gives the first real read.
We pulled every call above with one question to the Buzzberg MCP inside Claude: buzzberg.ai/mcp?utm_source…
0
0
7
619
Dylan Patel calls substrates the biggest bottleneck of all the bottlenecks
The tell: buyers now pay to build them.
Who's buying the Korean substrate maker, Samsung Electro-Mechanics, and why:
buzzberg.ai/ticker/009150.…
On Oct 21, Nvidia $NVDA said it is paying suppliers to grow substrate capacity 2.5x in 3 years, including a multibillion-dollar advance to Ibiden.
Samsung Electro-Mechanics went further: customer funding and volume guarantees for a ₩6.78T build in Korea and Vietnam.
Nvidia says its share will be a "significant portion" of the company's substrate output by 2028.
The catch: the new lines run from 2028-2030. Volume is promised. Price isn't.
So the question isn't whether there's a shortage. It's whether makers raise prices before the new capacity lands.
Investors we track, Oct 1-24: 14 buy, 0 bet against, 2 stay away.
First tell: Samsung Electro-Mechanics' Q3 results. Do they talk price, or only volume?
Our full Seoul Desk breakdown 👇
x.com/buzzberg_ai/st…
The tell: buyers now pay to build them.
Who's buying the Korean substrate maker, Samsung Electro-Mechanics, and why:
buzzberg.ai/ticker/009150.…
On Oct 21, Nvidia $NVDA said it is paying suppliers to grow substrate capacity 2.5x in 3 years, including a multibillion-dollar advance to Ibiden.
Samsung Electro-Mechanics went further: customer funding and volume guarantees for a ₩6.78T build in Korea and Vietnam.
Nvidia says its share will be a "significant portion" of the company's substrate output by 2028.
The catch: the new lines run from 2028-2030. Volume is promised. Price isn't.
So the question isn't whether there's a shortage. It's whether makers raise prices before the new capacity lands.
Investors we track, Oct 1-24: 14 buy, 0 bet against, 2 stay away.
First tell: Samsung Electro-Mechanics' Q3 results. Do they talk price, or only volume?
Our full Seoul Desk breakdown 👇
x.com/buzzberg_ai/st…
Dylan Patel explains why he thinks substrates are "the biggest bottleneck of all the bottlenecks," ahead of memory and optics, and bigger than power, land, or tools
“Yes, I think on the semiconductor side, the biggest constraint, of course, besides memory, is substrates, PCB terials, all of these components were severely under-invested in. Much of the supply chains that create the materials in Japan, those firms are sort of a lot more conservative and unwilling to invest heavily in capacity."
"And so we’ve seen, you know, tremendous bottlenecks there in terms of what can be made over the next two years, you know, both for accelerators as well as CPUs and every other component in the supply chain. So substrates is, I think, the biggest bottleneck of all the bottlenecks, including, you know, land, power, shell, semiconductors, tools."
"I’d say the next bottleneck would be, of course, memory. And then following that, there’s a lot of bottlenecks around the desire to ramp optics to a really, really large degree, right, with co-package optics or near-package optics. There’s humongous constraints in ramping that. You know, we just simply cannot ramp, you know, even if everyone wanted to do CPO in 2028, or even one major company wanted to do CPO in 2028, they could not satisfy their volumes in terms of lasers, in terms of fiber attach, in terms of all these different components that go into it.”
_______
Link and breakdown of the full panel with Dylan Patel and Stacy Rasgon: firesidealpha.substack.com/p/dylan-patel-…
0
0
6
602
Samsung Electro-Mechanics is spending ₩6.78T on chip substrates. Customers guaranteed the volume. The plan says nothing about price.
Every call we logged, and why 14 investors we track still say buy:
buzzberg.ai/ticker/009150.…
Seoul Desk: what Korean investors are saying
The plan: two new substrate factories (Sep 29 release)
- Sejong, Korea: ₩4.27T, its biggest single-product investment ever. Mass production starts Sep 2028
- Vietnam: about ₩2.51T, running by June 2030
- Customer funding and volume guarantees. No customer named. Nothing on price
So volume is promised, and price is the open question. It isn't the only thing that moves the stock, but it's the one this post looks at. Pricing helps determine what that promised volume earns.
The bull case: prices are already going up.
A Chesley Investment Advisory panel (Oct 22):
- Supply is so tight that makers are raising prices even on existing products
- This year's second-half increases come from scarcity, not from passing on costs
- Order backlogs jumped from the prior quarter
Even they said be picky: new 2027-28 capacity could squeeze margins if demand slows.
The cautious case: it has already run a long way.
Lee Ji-hwan, Aurora Investment Advisory (Oct 24), isn't betting against it. After its run, the stock trades at far higher multiples than Samsung Electronics, so in a pullback they would trim it and buy the chipmakers.
Investors we track on the stock, Oct 1-24: 14 buy, 0 bet against, 2 stay away.
How to read what comes next:
- Substrate prices still rising: support for the bull case, since the same volume earns more
- Prices flattening while the new lines are still being built: support for the cautious case
The first tell is the company's third-quarter results: do they talk about price, or only volume?
The bigger one comes later, in the fourth-quarter substrate price talks.
Every call we logged, and why 14 investors we track still say buy:
buzzberg.ai/ticker/009150.…
Seoul Desk: what Korean investors are saying
The plan: two new substrate factories (Sep 29 release)
- Sejong, Korea: ₩4.27T, its biggest single-product investment ever. Mass production starts Sep 2028
- Vietnam: about ₩2.51T, running by June 2030
- Customer funding and volume guarantees. No customer named. Nothing on price
So volume is promised, and price is the open question. It isn't the only thing that moves the stock, but it's the one this post looks at. Pricing helps determine what that promised volume earns.
The bull case: prices are already going up.
A Chesley Investment Advisory panel (Oct 22):
- Supply is so tight that makers are raising prices even on existing products
- This year's second-half increases come from scarcity, not from passing on costs
- Order backlogs jumped from the prior quarter
Even they said be picky: new 2027-28 capacity could squeeze margins if demand slows.
The cautious case: it has already run a long way.
Lee Ji-hwan, Aurora Investment Advisory (Oct 24), isn't betting against it. After its run, the stock trades at far higher multiples than Samsung Electronics, so in a pullback they would trim it and buy the chipmakers.
Investors we track on the stock, Oct 1-24: 14 buy, 0 bet against, 2 stay away.
How to read what comes next:
- Substrate prices still rising: support for the bull case, since the same volume earns more
- Prices flattening while the new lines are still being built: support for the cautious case
The first tell is the company's third-quarter results: do they talk about price, or only volume?
The bigger one comes later, in the fourth-quarter substrate price talks.
0
1
7
949
Jensen Huang’s AI bull case has a catch: customers need the money to keep buying.
On CNBC on Oct. 22, he said demand for AI cloud providers was off the charts.
The constraint was working capital. He expects financing to catch up.
He also rejected the suggested 2029 memory idth memory and LPDDR would be needed for years.
Two views worth following if you own AI stocks:
* Who can fund the next buildout
* How long memory demand lasts
Buzzberg brings @JensenHuang's dated views and original interviews into one profile, so you can follow his reasoning across the AI supply chain.
These are his public business views - not a disclosure of personal trades.
Which assumption does your AI position depend on? Open his profile and check the explanation behind it:
buzzberg.ai/speakers/jense…
On CNBC on Oct. 22, he said demand for AI cloud providers was off the charts.
The constraint was working capital. He expects financing to catch up.
He also rejected the suggested 2029 memory idth memory and LPDDR would be needed for years.
Two views worth following if you own AI stocks:
* Who can fund the next buildout
* How long memory demand lasts
Buzzberg brings @JensenHuang's dated views and original interviews into one profile, so you can follow his reasoning across the AI supply chain.
These are his public business views - not a disclosure of personal trades.
Which assumption does your AI position depend on? Open his profile and check the explanation behind it:
buzzberg.ai/speakers/jense…
1
1
9
703
One backed a biotech. One bought a token. One thought about protecting the portfolio.
The top 3 on Buzzberg’s X Top Calls board in our Oct. 23, 2026 snapshot. Prices have moved since.
🥇 @psyduckcapital - Silence Therapeutics $SLN: +116.0%
$9.10 → $19.66
He had discussed Oct. 7, ahead of ASH data. His Oct. 19 post revisited that earlier position; it wasn’t a fresh entry.
🥈 @ZoomerOracle - Lighter $LIT: +30.2%
$2.89 → $3.7622
He disclosed buying and explained his sizing through the number of tokens he wanted to own, using HYPE as a comparison.
🥉 @MoneyPrinter0x - $SOXL short: 21.4% move in the call’s favour
$104.48 → $82.11
On Oct. 21, he suggested the short as protection for a portfolio heavily exposed to risky crypto.
Two longs. One hedge. Credit to the speakers who put their thesis out!
These are tracked asset-price moves from the board’s recorded prices, not the speakers’ personal trading returns.
Which thesis would you still back at today’s price?
Open Top Calls and read the original posts behind the ranking:
buzzberg.ai/speakers?board…
The top 3 on Buzzberg’s X Top Calls board in our Oct. 23, 2026 snapshot. Prices have moved since.
🥇 @psyduckcapital - Silence Therapeutics $SLN: +116.0%
$9.10 → $19.66
He had discussed Oct. 7, ahead of ASH data. His Oct. 19 post revisited that earlier position; it wasn’t a fresh entry.
🥈 @ZoomerOracle - Lighter $LIT: +30.2%
$2.89 → $3.7622
He disclosed buying and explained his sizing through the number of tokens he wanted to own, using HYPE as a comparison.
🥉 @MoneyPrinter0x - $SOXL short: 21.4% move in the call’s favour
$104.48 → $82.11
On Oct. 21, he suggested the short as protection for a portfolio heavily exposed to risky crypto.
Two longs. One hedge. Credit to the speakers who put their thesis out!
These are tracked asset-price moves from the board’s recorded prices, not the speakers’ personal trading returns.
Which thesis would you still back at today’s price?
Open Top Calls and read the original posts behind the ranking:
buzzberg.ai/speakers?board…
1
0
7
626
Nvidia $NVDA: capacity, not demand, limits AI chips. On Oct 21 it paid suppliers to grow chip substrates 2.5x in 3 years.
Investors we track on substrate maker Samsung Electro-Mechanics (Oct 1-21): 12 buy, 0 bet against, 2 stay away.
Who saw it coming:
buzzberg.ai/ticker/009150.…
What a substrate does: it's the base an AI chip sits on, wiring the chip to the rest of the computer. Nvidia can't ship a chip without one.
Who's building what for Nvidia (its Oct 21 release):
• Ibiden (Japan, substrates): a multibillion-dollar advance, its Ono plant to 3x, from early 2028
• Samsung Electro-Mechanics (Korea, substrates): new to Nvidia's supply chain, with new or upgraded plants in Sejong and Busan
• Unimicron (Taiwan, substrates): more capacity from April 2027, all of it by mid-2028
• Amkor $AMKR (packaging): interposer capacity up more than 300% by end-2028
• KYEC (Taiwan, chip testing): a $1.1B advance for new US test sites, from April 2027
@jensenhuang: "Demand is extraordinary. Capacity is the constraint."
Two investors we track made the buy case before the news:
• Noh Geun-chang (815 Money Talk), Oct 20: substrate profits were set to turn up, with customers helping pay for the build
• Yoon Ji-ho (#15 on our 30-day board, 3PRO TV), Oct 16: buy the companies with the most growth by 2028
Nvidia's release says its share will be a "significant portion" of Samsung Electro-Mechanics' substrate output by 2028.
What would change this: these plants come online in 2027-2028. If AI spending slows before then, the new capacity could arrive into weaker demand.
Which supplier gets the most out of this?
Investors we track on substrate maker Samsung Electro-Mechanics (Oct 1-21): 12 buy, 0 bet against, 2 stay away.
Who saw it coming:
buzzberg.ai/ticker/009150.…
What a substrate does: it's the base an AI chip sits on, wiring the chip to the rest of the computer. Nvidia can't ship a chip without one.
Who's building what for Nvidia (its Oct 21 release):
• Ibiden (Japan, substrates): a multibillion-dollar advance, its Ono plant to 3x, from early 2028
• Samsung Electro-Mechanics (Korea, substrates): new to Nvidia's supply chain, with new or upgraded plants in Sejong and Busan
• Unimicron (Taiwan, substrates): more capacity from April 2027, all of it by mid-2028
• Amkor $AMKR (packaging): interposer capacity up more than 300% by end-2028
• KYEC (Taiwan, chip testing): a $1.1B advance for new US test sites, from April 2027
@jensenhuang: "Demand is extraordinary. Capacity is the constraint."
Two investors we track made the buy case before the news:
• Noh Geun-chang (815 Money Talk), Oct 20: substrate profits were set to turn up, with customers helping pay for the build
• Yoon Ji-ho (#15 on our 30-day board, 3PRO TV), Oct 16: buy the companies with the most growth by 2028
Nvidia's release says its share will be a "significant portion" of Samsung Electro-Mechanics' substrate output by 2028.
What would change this: these plants come online in 2027-2028. If AI spending slows before then, the new capacity could arrive into weaker demand.
Which supplier gets the most out of this?
0
0
6
1,236