Max Song (๐, ๐ , ๐ ) retweeted
If AI had been writing and reviewing its own code for the past few decades, and humans had only recently started looking into what it was up to and discovered an ever-growing list of CVEs, we would reasonably conclude that AI hadnโt been good at coding all along and that we should just start over.
Except itโs the other way around.
Humans have been the ones writing and reviewing each otherโs code for decades. Very diligently, if I may add.
If weโre being intellectually honest and holding ourselves to the same bar we hold AI to, weโd have to admit that we havenโt been all that good at it either.
Max Song (๐, ๐ , ๐ ) retweeted
So basically, today PHP got two critical bugs that let anyone run code on your server without even logging in, which is bad, and then the details leaked before the patch was even out, which is worse, and it works on WordPress and MediaWiki and Magento and Joomla and Laravel, and Censys says like 1.5 million servers are running a version with no fix, so the NSA was like "take your system offline," and the FBI was like "take your system offline," and CISA and the UK and CERT-EU were all like "take your system offline," and then the Prime Minister of Japan, the actual Prime Minister, posted at 5:55 in the morning telling 48.8 million people to take their system offline, so everyone took their system offline, like the Dutch cybersecurity agency took its own website offline, and the Dutch parliament went offline, and the Port of Rotterdam went offline, and Finland and France and Denmark started pulling government sites, and Wikipedia went down, and Archive.org went down, and Asana and GOG and Something Awful went down, and the University of Vienna's webmail went down for some reason, and that's just PHP. Because GitLab also dropped a CVSS 10.0, which is the highest score they give out, for a bug that's been there since 2014, so that's every self-hosted GitLab for twelve years, and then LiteLLM 1.107.0 got hijacked and shipped a credential stealer and a remote access trojan in every single wheel and the Docker images, including latest, and you didn't even have to import it, just installing it was enough, and it's the second time this year, and then Railway paused every deployment on the entire platform because someone got into the environment variables, so now everyone has to rotate every secret they've ever stored there, and TiDB Cloud said someone was walking around inside with stolen credentials for basically a month and an extortion group says they took 10.6 terabytes, and Zeabur got breached for the second time in two months and is just shutting down, like the whole company, and Node.js patched remote code execution and a TLS bypass, and CISA says Plex is being actively exploited, and someone took over 300 Tor exit relays to steal SSH passwords, and SafePay ransomware has hit over 600 critical infrastructure orgs, and India's biggest bank filed a data loss incident, and Japan has been in an official state of emergency in cyberspace since October 9, which honestly explains the 5:55 AM post. Oh, and eMudhra, a certificate authority that's issued over 60 million certificates, whose entire job is proving websites are who they say they are, was supposed to be using private keys that are random numbers over a hundred digits long, and one of them was 4. Just 4. And the others were like 66 and 237. Anyway, it's Wednesday.
Max Song (๐, ๐ , ๐ ) retweeted
what an absolutely insane chart.
8 companies will spend $1.5+ trillion in 2027 and $690 billion of that will go into nvidia's pockets. thats 44 cents out of every dollar lol.
google is the largest spender at $375 billion on TPUs, microsoft in 2nd. coincidence that these 2 companies fell behind on frontier models?
pretty clear that largest spend does not equal best model.
overall this is the largest expenditure every witnessed in history and its now forcing neoclouds to go cash-flow negative.
Max Song (๐, ๐ , ๐ ) retweeted
Anthropicโs report is basically saying humans are the bottleneck.
โAmid this model-driven acceleration, human attention is a constant bottleneck. Claude sessions amount to around 80% idly waiting for human input. As one engineer put it, people are โbottlenecking the agents.โ
Max Song (๐, ๐ , ๐ ) retweeted
When I was writing Unreal Engine 1, my time was divided as follows:
- Around 30% went to seriously deep coding on hard problems: assembly language texture mappers, C++ BSP trees, data persistence, object oriented systems design, and networking.
- Around 70% was just typing and refactoring large bodies of code needed for a complete engine feature set. This was not hard, and while I was doing it my mind was mostly on the deeper coding topics, thinking and planning for what's next. It didn't feel much different from vibe coding today, except that it was just me typing code on an IBM Model M keyboard rather than prompting it and reviewing the results.
So, it has always seemed perfectly logical that we would rely on AI for that 70%, once the AI could do it with a similar quality standard as we could, which has never been perfection. It came as no surprise that, after a few years of AI model evolution, it was able to take the reigns on that 70% and free up programmers' time for more valuable tasks.
What's more of a surprise is the emergence of AI as a sort of insta-Carmack able to solve hard technical problems quite cleverly and often cleanly, and that it's able to reason about and work in complex 3D scenes both by operating the tools and by directly understanding the raw data files.
I abandoned assembly language in 1998 when out-of-order CPUs made much simpler C/C++ code faster, and never I missed it. If prompting through architectural documents, concept art, and design plans can transcend typing code and tweaking vertices, that will enable developers to spend their time on higher-level, higher value technical and creative tasks.
What matters is the end product, whether it's a game or an engine. Anything developers can do to try out more ideas and make better software and games faster is a win for everyone.
Max Song (๐, ๐ , ๐ ) retweeted
In the last year or so, in talks & podcasts, I predicted this AI-bio-autolab-digital twin loop as a key pillar for solving all diseases & reversing aging in 10-15 years. โExpertsโ called it unrealistic & impossible in such a short time. Turns out I may have been conservative.
We are bringing agents to the wet-lab bench at Google Research to navigate complex biological landscapes. Learn how we are grounding agents in physical lab robotics to run fast closed-loop molecular design cycles, ultimately discovering single-digit nanomolar protein binders for cancer targets โgoo.gle/4B5mMMB
Max Song (๐, ๐ , ๐ ) retweeted
LLMs are either not capable or not affordable for visual reasoning tasks at scale.
The Elorian team - one of Jeff Deanโs best reports who led Gemini, an Apple LLM lead, xAI lead, and a Harvard professor - just launched an Astra-level visual reasoning model, Glimpse, for 17x cheaper.
Watch it break down the flying Tesla video physics here.
Extremely critical for video inputs in manufacturing, robotics, gaming, content moderation and more.
Max Song (๐, ๐ , ๐ ) retweeted
CLAUDE FOUND THE SEEDS
anthropic.com/research/ecdsaโฆ
Awesome essay and philosophy
This might be the most important essay of 2026.
My conversations with Nick over the past few months about his research have given me great hope for the future. I'm so happy he is sharing these findings with the world.
AI presents extremely difficult problems. We can all assume these problems are intractable and that we are doomed, or we can try and solve for a positive future.
Personally, I choose optimism. There are many challenges ahead... let us build the tools to face them head on.
How long do we have to actual talk to our doctors?? Not enough by any means. This is a wonderful study
Your doctor cuts you off about 11 seconds after you start talking. Researchers literally timed it in a 2018 study. That number is why Google's new Lancet result is bigger than it looks.
In the study, 98 real urgent care patients texted with Google's AI days before their appointment. No scans, no exam, no bloodwork, just an unhurried conversation about what was wrong. That alone put the correct final diagnosis on the AI's list 90% of the time.
Sounds like magic until you learn what medicine has known for 50 years.
Back in 1975, a British team published a study in the BMJ tracking which part of a visit actually produced the right diagnosis. The physical exam? The labs? Nope. In 82% of cases, the patient's story alone got the doctor there. Everything else mostly confirmed what listening had already found.
So the playbook for accurate diagnosis has existed since 1975. Let the patient talk. Medicine just built a system where nobody has time to run it, and an 18-minute appointment with 11 seconds of uninterrupted patient talk was the result.
Google didn't build a superhuman doctor. It built the first clinician in 50 years with enough time to use the method we already proved works.
And my favorite detail buried in the study... physicians who read the AI's pre-visit summary said it changed how they treated the patient in over half of cases. Hand a doctor the full story before you walk in the room and you get different medicine.
Turns out the most advanced medical instrument Google ever shipped is a conversation that doesn't get interrupted.
Max Song (๐, ๐ , ๐ ) retweeted
Allegedly โthe most important chartโ for understanding the current state of the AI economy. The model layer seems to be going through a period of commodification as non-frontier models saturate everyday use while frontier models compete fiercely. Compute layer sees massive increase in total demand for compute regardless of the source or use case.
Max Song (๐, ๐ , ๐ ) retweeted
Today we announced interim data from our first-in-human, Phase 1 clinical trial evaluating our lead program in optic neuropathies, showing it was well-tolerated by 3 participants with glaucoma.
Read more about our safety & visual field observations here:bit.ly/4ASWq0p
Max Song (๐, ๐ , ๐ ) retweeted
David Vรฉlez is the founder of Nubank, a bank with 140 million customers and a market capitalization of $60 billion.
When he started the business in 2013, no one thought it would work, not even his investors. He has faced down regulators, oligopolies, and corruption to build Latin America's largest digital bank. Now he is bringing Nubank to the United States, where he faces his biggest challenge yet.
@gabi_imarques, who grew up four miles from Nubank's headquarters in Brazil, spent time with @velez_david in Sรฃo Paulo, where the business started, and in Miami, its latest beachhead.
She profiles a man who is no stranger to a fight, having grown up in Medellรญn, Colombia, when the city was at war with Pablo Escobar. Six-year-old Vรฉlez would pull on layers of sweaters before stepping outside, hoping, as a child might, that the extra padding would cushion any bullets.
colossus.com/article/david-vโฆ
Max Song (๐, ๐ , ๐ ) retweeted
For folks not understanding where we are at with AI now - if the task is verifiable, it is now solved by AI.
Yes, earlier days you could argue that 'AI isn't that good' (e.g. see first attached chart here, where GPT-4o underperformed average accountants on tasks)
But now, it's a totally different story (only ~2 years later). The second chart is astounding. It took me a while to even understand it because it looks so odd. It compares manual accounting tasks to tasks complete with Opus 5.5. Basically Opus 5.5. solves all tasks almost instantly at a 100% accuracy rate, whereas a person takes a lot more time and gets a lot of things wrong. It makes the chart look entirely silly because the axes aren't even comparable.
So yeah, this is where we're at. If it is verifiable, it is solved. This is just how these model architectures work now.
Max Song (๐, ๐ , ๐ ) retweeted
New research: we hired 12 CPAs to complete four realistic accounting tasks, then had AI models from the past few years complete the same work.
As of May 2025, accountants outperformed AI. Now they lose out to today's frontier models, which are near perfect on these tasks. ๐งต
Max Song (๐, ๐ , ๐ ) retweeted
Where does money invested into the AI buildout actually go?
For every $100 flowing into the supply chain:
- $50 to chips
- $20 to power
- $15 to networking
- $15 to cooling, buildings, and land
More charts in State of Markets II: a16z.news/p/state-of-marketsโฆ
Introducing our State of Markets pt 2, along with a companion podcast where we unpack the data and discuss what comes next.
Tech is the everything cycle.
Supply: putting the buildout in context, just passed railroads as % of GDP. The wisdom of Elon is real: the factory (or the datacenter!) is the product.
Demand: diffusion is so, so early. Median AI vendor spending in the top 1% of companies is 8x that of the top 10%. Only about 30% of S&P 500 companies report a quantified AI impact, which means thereโs a substantial opportunity in connecting models to a company's data and workflows. Diffusion into companies is one of the main themes of the next 5 years.
Weโre entering the agent work period. Only a few million users today, but applicable to billions of internet users with massive surplus created. META/GOOG monetize US users at $200+ per year today. Agent opportunity is much higher.
Mega-trends the next 5 years: Consumer agents, Robotics, Autonomy, AI x bio, Personal health, Diffusion into enterprise, New era of American Dynamism.
Much more in our SoM report here - a16z.news/p/state-of-marketsโฆ
@a16z @sarahdingwang @aleximm @santiago__rdz
Max Song (๐, ๐ , ๐ ) retweeted
98% of US households aren't paying for AI yet
More charts in State of Markets II: a16z.news/p/state-of-marketsโฆ
Introducing our State of Markets pt 2, along with a companion podcast where we unpack the data and discuss what comes next.
Tech is the everything cycle.
Supply: putting the buildout in context, just passed railroads as % of GDP. The wisdom of Elon is real: the factory (or the datacenter!) is the product.
Demand: diffusion is so, so early. Median AI vendor spending in the top 1% of companies is 8x that of the top 10%. Only about 30% of S&P 500 companies report a quantified AI impact, which means thereโs a substantial opportunity in connecting models to a company's data and workflows. Diffusion into companies is one of the main themes of the next 5 years.
Weโre entering the agent work period. Only a few million users today, but applicable to billions of internet users with massive surplus created. META/GOOG monetize US users at $200+ per year today. Agent opportunity is much higher.
Mega-trends the next 5 years: Consumer agents, Robotics, Autonomy, AI x bio, Personal health, Diffusion into enterprise, New era of American Dynamism.
Much more in our SoM report here - a16z.news/p/state-of-marketsโฆ
@a16z @sarahdingwang @aleximm @santiago__rdz
Massive data point for AI Education
Researchers proved AI has deleted every reason universities exist.
Harvard University ran a controlled experiment pitting a custom AI against their own top-tier classrooms.
And the results are going to collapse the higher education bubble.
They took 194 undergraduates and split them up.
One group learned physics in one of Harvardโs best hands-on, active-learning physical classrooms. Group work. Instructor support. The premium university experience.
The other group went home and learned the exact same material with an AI tutor.
The AI didn't just win. It embarrassed the institution.
Students using the AI learned more than twice as much as the students in the elite Harvard classroom.
They scored 30% higher on the final assessment.
And they did it in less time.
Let that sink in.
A piece of software sitting on a laptop outperformed a world-class faculty in one of the most elite learning environments on Earth.
Universities have always justified their exorbitant tuition with two things: access to elite knowledge and the physical classroom experience.
This study just proved both of those moats are gone.
When software can teach you complex physics twice as well as a $60,000-a-year institution, the math of higher education breaks permanently.
The AI didn't just give the students answers. It used strict pedagogical guardrails. It guided. It questioned. It forced the students to do the cognitive work.
It offered perfect, one-to-one tutoring, personalized to the exact moment a student misunderstood a concept.
That level of attention is mathematically impossible to scale in a physical lecture hall.
For a thousand years, the university was the only place to get a premium education.
Now, itโs the bottleneck.
If AI can double your learning speed for a fraction of the cost, what exactly are students taking on decades of debt to pay for?
