can we still talk about john ternus? cause srsly apple picking a hardware lifer over a services or marketing guy tells you exactly what they think their next decade is about โ building things, not selling subscriptions...
@Sam_Badawi this fits the pattern โ they went from a $2.5b facility last year to $15b now, right as the confidential S-1 turns public. the credit line isn't the story, it's the tell that the ipo clock actually started
1) Every day, I get more and more convinced that AI is nowhere near saturation levels as AI will continue to do more digitally, then take on autonomous driving and then a lot more things with humanoid robotics.
2) I want to invest based on assumption #1 being correct...so I get excited for semiconductors (like $NVDA, $AMD, $MU, etc.), AI model providers ($GOOGL, $SPCX, $META, OpenAI and Anthropic), and infrastructure/power providers ($CRWV, $IREN, $EOSE, etc.).
3) As a long-term investor, I want to balance risk/reward, and I know the importance of valuation. IF the thesis is wrong (#1), the hardest hit companies will be those that are most over-valued..so I look to see which of these companies have the best risk/reward. See the attached table summarizing things.
4) From this table, there are clearly 4 companies that stand out: $NVDA, $MU, $META and $DELL. These companies all have Forward PEs that are below the S&P 500 average, which is fascinating to me considering they are all high-growing tech companies.
5) Of the 4 companies, 3 of them ($NVDA, $MU and $DELL) are directly tied to AI & Data center hardware. Within this group of 3, there is an outlier with forward P/E of ~6.7 vs. ~18 for the others. While 18 is good, 6 is UNBELIEVABLE!
And the 4th company, $META, which has a similar PE to $NVDA and $DELL, has no current dependency on AI/Data Center spend. In fact, just the opposite. Meta would become MORE profitable if it stopped spending on AI & Data Centers.
6) Therefore, I'm heavily invested in just 2 of these names:
$MU, which is now ~33% of my portfolio
$META, which is now ~17% of my portfolio
50% of my portfolio revolves around AI growth, but it's focused on just 2 companies that I believe to be the most under-valued with the best risk/reward.
@higgsfield_ai this is a good example of what's actually changing right now โ not "ai makes better images" but ai as one link in a real production pipeline (code gen > 3d build > render) that used to need a whole team...
@perplexity_ai we're at the point where "highest score AND cheaper" is the actual battleground, not just raw benchmark numbers. price-per-task is quietly becoming the real leaderboard
sooo john ternus just became apple's ceo (inheriting the only big tech company without a frontier model of its own), and openai rated its next model a "critical cyber risk"
this is the first time a lab has self-rated a model that seriously and kept building anyway. the actual story isn't the model, it's that the safety framework is now just... documentation
srsly evry agent company is racing to solve the wrong problem. the blocker was never capability, it's auth. nobody's cracked "give an agent access without giving it everything"
Claude can now use your computer in the background in Claude Cowork and Claude Code.
Give it something to do on your desktop and Claude clicks, types, and opens apps just like you would, while you work on something else.
this is the actual bottleneck for agentic browsing rn, not capability. until there's a real auth layer that isn't "paste your password into a vm," adoption caps out at people who don't think about security
I am a big fan of Grok @bot, and itโs clear that other personal agents are moving toward using a persistent cloud computer.
But thereโs just something that feels wrong about entering my password and authentication code into a cloud browser.
Beyond supporting more official plugins, Iโm not sure there are easy solutions to this. I talk more about it in my latest video here: https://t.co/Y2ez5Wm7kq
@StockSavvyShay lemonade pricing FSD miles 50% below regular driving is such a fun little data point. insurance quietly becoming the tesla autonomy scoreboard...
@KyleReidhead $NVDA thinks physical ai will be bigger than the chatgpt moment. robotaxis first, humanoids next, basically evry company becomes a robotics company eventually!
@maya_movess hnstly love that companies are finally realizing not every task needs the fanciest model, matching workflows to the right model is such a smart way to actually save money without losing quality!!
meta's updating its ai glasses to stop recording if someone covers the capture led, cracking down on covert filming and harassment!! meta's ai glasses finally getting a "no you can't secretly film people" update is honestly such an overdue fix, glad accountability caught up to the tech eventually ๐ค
I wonder if when they say โgood models take timeโ they were referring to GPT-OSS-2T for dev day.
Everyone thought it was Astra that was coming on dev day, however that was never the case.
a new open source model on dev day would be a great surprise.
@KyleReidhead loving the vision here hnstly, tokenized dividends auto-swapping into whatever u want is the kind of flexibility brokerages have never given retail, worth watching if adoption actually catches up to the idea...
@cb_doge "windows downloads 70% smaller" is such a small line for something that's going to make so many people's lives easier, the unglamorous fixes are always the ones that matter most hnstly