I’m a contrarian, so often I see the world differently than traditional investors.
I’ll put my Board member hat on here and think like the Chair of a Strategic Transactions Committee.
🧵 👇
Hot take: Universities charge $300,000 for a degree that teaches you skills any LLM can do for free. At some point we need to have an honest conversation about whether higher education is the greatest individuals misallocation of capital in recent history.
One of the world's most analytically demanding hedge funds replaced junior analyst drafting time with AI. Not a pilot. In production.
@BillAckman told @theallinpod that Pershing Square uses AI to write first drafts of investment documents. Small team, concentrated bets, highest standards. Already made the transition.
The hours junior analysts spent on initial memos and research summaries are gone. Same team. More output.
If Pershing Square has already done this, every firm that hasn't is running with a structural cost disadvantage.
This is not a 2027 problem: https://t.co/3axkkQIOmU
Source: All-In Podcast - https://t.co/3EnRCmQ2DA
College is extraordinarily expensive and becoming less useful, and those who insist otherwise are working from a model of the labor market that stopped describing reality sometime in the 1990s. Four-year courses at private institutions often cost more than $70,000 a year, and it should come as no surprise that student debt has tipped over $1 trillion .
✍️ @default_friend
BREAKING: Kirkland & Ellis and Palantir agreed to a multiyear deal to develop AI technology to help it advise private equity groups on raising money from investors like pension funds.
The Palantir tool would be used for fund documentation, draft side letters, and keep track of PE firms’ agreements with investors.
@NYMag Massive gap in price vs value
These tuitions are a fraud for all but the best universities
The university product is weak and steeply overpriced
Huge opportunity for disruption
AI is the catalyst
@zero_university points the way
Here’s the formula enterprises are going to use to keep AI + agent bills under control
Ranked from easiest to hardest:
> model routing
Obvious first step, don’t send every task to the most expensive frontier model
Some companies will use 3rd party routers, others will build routing logic in-house
> prompt + context trimming
Most users will default to dumping everything into the most capable model, super inefficient
Companies will build a layer between the user and the model that decides:
What context is actually needed?
What can be retrieved later?
What can be summarized?
What can be cached?
What level of reasoning is needed?
What model should receive it?
> custom agent harnesses
Enterprises will not just let agents run wild, they will build or buy harnesses that manage:
tool access
memory
permissions
workflow state
evals
logging
human approvals
fallbacks
cost controls
Some will use external products like Hermes Agent + OpenShell, some will build their own internal harnesses
> heavy use of open-weight models + SFT on proprietary data
This is the hardest but potentially the most powerful
For high-volume repeatable workflows, companies will increasingly use open-weight models, fine-tuned on their own data, running on cheaper dedicated infra
Not every task needs the smartest model in the world, many just need a model tuned for the company’s use cases
All of the above will be wrapped with governance and budgets, the enterprise AI stack will optimize for intelligence per dollar
Sounds impressive, well done
Marketplaces will develop for such skills libraries across disciplines
Instantiating process expertise in skills files will be commoditized over time IMHO, or they will be guarded like state secrets, i.e. Coke’s recipe
To call them the most sophisticated implies that your underlying skills in analytics are the most sophisticated as well as your instantiation of such skills in skills libraries
Perhaps it’s both
Seems that the process of investing for managers will be improved through leveraging best in class skills libraries
What happens when the best in class skills libraries are table stakes for managers?
Complete waste of time imo to do actual fundamental analysis now. You should be thinking about how these tools will cause others to make mistakes, flows, positioning etc. all your effort should go into the Keynesian beauty contest, the fundamental story can easily be digested in 5 minutes now with little juice left to squeeze by doing more work.
Scoop! Lila Sciences is in talks to raise ~$2 billion in new funding.
The raise would value the “scientific superintelligence” lab at ~$8.5 billion before the new money.
w/ @MichelleF_Davis, read more👇
In 712 Harvard courses last year, every undergrad got an A. The Crimson analyzed a decade of confidential grading data to map where the grading cap — set to take effect in Fall 2027 — will hit hardest.
@abbysgerstein and @amannmahajan report.
https://t.co/zOn9rgK6vw
@McGillPatterson Understood
For background, I have been closely following the ed tech space for over 30 years and have five children for whom I have paid university tuition and experienced the product
Law professors overwhelmingly preferred answers drafted by AI over ones written by fellow professors, a new Stanford Law School study found, suggesting that the technology is capable of legal reasoning and that law students may benefit from AI tutoring. https://t.co/NFBOrpceUB
This is a real and immediate issue
Failing to embrace AI leaves graduates ill prepared for the workplace
One advantage of a university is the time afforded students to learn
Graduating a student without fluency in AI is a dereliction of duty for universities
The gap between value and price of higher education is wide
@McGillPatterson The education establishment has never had a competitive source of delivering learning
AI provides that today
The higher ed stack is going to become unbundled and repriced
This horse is out of the barn
Common core, grade inflation, parents who want their children coddled and rewarded for participation have all contributed
And universities have been happy to accept this product and pass it along to the workplace, capturing ever larger tuitions and donations in the process
The education system is bankrupt
“The baseline for what students consider a reasonable amount of work has dropped so noticeably that he has cut his readings accordingly; a 750-word essay now strikes many students as long” https://t.co/CEzkDq5ZYN
🔥🚨BREAKING: It is being claimed on X that In 2009, Tim Tebow proved the existence of God, a post that is trending reveals the statistically impossible consistency with a bible verse and the former football player.
Trad West: ‘Tim Tebow wrote John 3:16” under his eyes during the National Championship Game. 94 million people searched the verse that night.
Exactly three years later to the day, Tebow played his first NFL playoff game against the Pittsburgh Steelers.
He threw for 316 yards. His yards per completion were 31.6. The TV rating peaked at 31.6. The opponent’s time of possession was 31:06. The only interception in the game was thrown on 3rd and 16. The game was played exactly 316 weeks after Tebow declared he would play college football for the University of Florida.
Six different statistics. All pointing to the same verse. On the same date. Three years apart.
Nobody planned this. Nobody could have.’
“For God so loved the world that he gave his one and only Son, that whoever believes in him shall not perish but have eternal life.” John 3:16