
Learn something new every day. Some days this is easy; some days, you have to claw for it. Do it anyway. To learn is to grow.

Computer scientist+MBA, pilot, inglés+español. @Java_Champions, ⭐️presenter, author @springbootbook. Ex @Oracle, @Pivotal, @VMware, @Microsoft

Learn something new every day. Some days this is easy; some days, you have to claw for it. Do it anyway. To learn is to grow.

“If an opinion contrary to your own makes you angry, that is a sign that you are subconsciously aware of having no good reason for thinking as you do.” ~ Bertrand Russell Philosopher, logician, mathematician, social critic, winner of the Nobel Prize in literature #qotd

Grudgingly awarding points for the reference. 🖖😐
Java4Shell: write once, RCE everywhere calif.io/java4shell

With all the outages happening, it's hard to not lean towards something larger happening. We still have no evidence of mass outages caused due to cyber. The FBI warning around ARTEX was more of an advisory around it being used by threat actors doing active campaigns, not disruption efforts. Most folks don't realize that our interconnected systems, predominantly because we put all of our eggs in large baskets causes much of this. Instability in large cloud providers cause not one outage, but thousands if not hundreds/millions of outages as we saw last week with Microsoft. I know folks in almost every organization that has been impacted here, folks I've known for years - all confirming it was not cyber-related. In stating that - is it possible? Absolutely. However, a lot of what we are seeing is hyper sensitivity - any blip is going to cause a reaction right now. Again, can't rule out cyber attacks in any stretch, and I'm deeply concerned on what we are seeing across the spectrum from a security operations perspective and incident respond side. It is definitely elevating upward for data breaches especially with autonomous attacks. The next 6 months are going to be rough, at the same time - we will always find a way to get better, improve, make these less impactful. I want to emphasize again, none of these outages have been confirmed a cyber attack and have clear explanations on what has happened. Which honestly, should concern us more. It's showing how weak our infrastructure in supporting our daily lives really is, and how that can change very quickly. We must do better.

🚨 BREAKING: A critical Java flaw that reportedly lets hackers hijack servers without a password leaked before its patch. Security firm Calif says it found the bug, dubbed Java4Shell, and will publish details today. Oracle fixes 11 Java flaws today but hasn't tied any to it.

The #FBI issued a FLASH advisory warning that cybercriminals and state-aligned hackers are exploiting ARTEX, a super intelligence-powered penetration testing framework, to automate attacks against businesses and organizations worldwide. ARTEX can accelerate intrusions from initial targeting to privileged access in under three hours. It was used in the October ConnectWise supply-chain compromise, which exposed Microsoft 365 customer environments and involved attempted authentication against more than 2,250 downstream tenants. The FBI is urging businesses and organizations to strengthen access controls, monitor suspicious automated activity, and report potential compromises to ic3.gov. Read more here: ic3.gov/CSA/2026/261113.pdf

🦔Hackers planted card-stealing code on the websites of 178 companies on Monday, including Zoom, CNN, the BBC, IKEA and HSBC. They got in through Optimizely, a website tool all of those companies use, and the malware went out to their visitors. My Take Optimizely’s tool exists to test which headline or button color gets more clicks. Thirty-two banking and finance sites let it run code on their pages anyway, and I have no idea why a bank would let a marketing tool anywhere near the page where you type your card number. The attacker went through a vendor HSBC trusted and never had to touch HSBC itself (rough week for HSBC). If you used one of these sites Monday, update your iPhone or iPad (the code went after older versions) and check your card statement. Anyone who sent crypto that day should double-check the address it went to. Hedgie🤗

Tomorrow night! I'll be talking to Miami JUG about AI, agents, and MCP. Should be a fun time, so I'd love to (virtually) see you there! 👩💻 #neo4j #AI #agents #mcp meetup.com/miami-java-user-g…

A film you've watched more than six times with a gif. No Star Trek, Star Wars, Lord of the Rings, Marvel, Disney or Ghibli. Either this or any Disney/Pixar movie ever since becoming a parent 🤣
A film you've watched more than six times with a gif. No Star Trek, Star Wars, Lord of the Rings, Marvel, Disney or Ghibli. Either this or any Disney/Pixar movie ever since becoming a parent 🤣

I'm signal-boosting this because Luke clearly knows what he's talking about. I think his diagnosis that Microsoft's running cloud services are only metastable and cannot handle the restart storm is almost certainly correct.
Lots of hot takes on the Microsoft 365 outage. Here's an engineering one. To be clear: this is speculation. Microsoft hasn't confirmed a root cause. It's my read of their own updates, from 35 years of building and breaking systems. What Microsoft has actually said: • Traffic isn't up. Capacity is down. Servers are failing health checks and dropping out. • A process that establishes secure connections gets overwhelmed, and servers stop processing requests. • Servers they bring back fall over again as soon as traffic hits them. • They're reverting a network config change that lines up with when it started. But changes aren't syncing, and some machines are stuck in a bad state. • No indication of an attack. My hypothesis: a metastable failure. Whatever triggered it may be long gone. The outage is now keeping itself alive. 1. Reconnect storm. Every Outlook, Teams and mobile client on the planet is hammering for a fresh secure connection. Full TLS handshakes are CPU-heavy, and a freshly rebooted server has a cold session cache. A server comes up, eats a wall of handshakes, CPU pegs, it fails its health check and gets pulled, and its load lands on the next one. Dominoes. 2. The fix shares fate with the failure. If the rollback travels over the same secure channel that's melting, the cure can't reach the patient. That's exactly what "changes aren't syncing" sounds like to me. That's why they're throttling traffic types and rebuilding servers from an older config instead of flipping a switch. You don't get out of this by finding the bug. You get out by shedding load, restoring capacity in isolated cells, and letting traffic back in slowly. Where AI fits (also speculation): not "an AI wrote the bad line". It's velocity. Microsoft flagged several recent config changes as suspects and took hours to find one that correlated. When the volume of change outpaces the humans who understand each change, finding the culprit gets slow and rollbacks get messy. Automate the shipping faster than you automate the understanding, and this is the bill. The takeaway for every business: your email, chat, files, identity and security tooling now share one failure domain. Today proved it. Have a way to talk to your team that doesn't live in the same tenant. Happy to be proven wrong by the post-incident review. And I hope this post ages terribly.

I have a major result to share. I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China. This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model. Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves. I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems. I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used. It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20. These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs. I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at unemploymentlabs.com/checkpo…. There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

🦔 Jensen Huang resigned as Nvidia’s CEO this morning, four days after the board said it fully backed him. CFO Colette Kress runs the company until the board picks a replacement. My Take Huang was the AI trade’s best salesman, and a lot of Nvidia’s value rested on investors trusting his forecasts. Kress is a numbers person, and numbers people don’t stand on a keynote stage and promise trillions in AI spending. So even if Tuesday’s quarter looks great, I see the stock losing some of its “Jensen premium” for good. The bigger risk is downstream. He kept the circular money flowing into OpenAI, CoreWeave and the neoclouds, and whoever gets this job permanently won’t feel his loyalty to those bets. If that cash slows, the weaker AI players head to private credit at worse rates, and that’s where this boom has always looked most fragile to me. Hedgie🤗

🦔A longtime Broadcom executive says the company fired him to avoid paying out more than $5 million in stock. Eitan Cohen, 58, spent 18 years at Broadcom and ran a chip development team of about 360 people in Bulgaria. Broadcom let him go in January as part of a workforce cut. He wants about $1 million in damages from a Tel Aviv labor court and also alleges age discrimination. Broadcom denies all of it and says the decision came down to business needs. My Take I can't say who's right here, and the court will sort that out. Plenty of people with stock pay don't know how fragile it is, though. Restricted stock units only become yours once they vest, usually a slice each year, and if you lose your job before then the unvested shares typically disappear. Broadcom says that's standard, and it is. That setup gives companies an easy way to save money in a layoff. Cut the people with the biggest unvested piles and the stock goes right back to the company. I'd like to see partly earned stock pay out on a prorated basis when a company lets someone go without cause. Until that happens, I wouldn't count unvested shares toward a budget or a down payment. Hedgie🤗

It looks like the AI layoff boom has started to backfire. I was just going through a Reuters investigation on Meta, and holy shit... They drew up plans to cut some teams by as much as 60% and let AI agents do the work. THEN Zuckerberg called off the next round of layoffs hours before the first one even started. And when you look at the numbers... they tell you why. Engineers were pushing out 220% more code, BUT major technical and security incidents jumped 40%. So people spent 70% more time putting out fires. AI cranked out way more work, and humans spent way more time cleaning it up. A different report I saw this week showed something similar. 32% of hiring managers said they brought back the exact jobs they cut for AI. Think about how dumb that is. You fire people, pay a fortune for AI, then pay again to hire the same people back. Look, I’m no AI bear. I use AI every single day. It saves me a shit ton of time and I’m not going back. It’s an incredible tool. But market forces are still market forces. Every boom in history has been overhyped, and every CEO wanted to stand up and tell Wall Street he could do more with half the people. Now from what I’m seeing, reality is kicking in. It costs way more than they thought, it breaks shit, and you still need people who know what the hell they’re doing. Even the AI companies are feeling it. Elon just cut Grok’s usage limits and admitted it costs more to run than it brings in. I still think AI is a great bet, and I’m still buying the companies building the data centers. But you need balance, and the guys who went all in on firing everyone are learning that the hard way. Be careful out there.

Folks, I’m growing the DX team at Render! I’m hiring 4 super talented people to build cool projects and get developers (and their agents) excited to ship on Render too. The roles: - Developer programs lead - DX Engineer - Developer Relations Engineer - Content Engineer. Remote or in SF office. Links in 🧵

The web rewrite is working. On a slow connection, the old X web took more than 22 seconds to show the first post on a profile. The new X web does it in 2.7 seconds. Here's how we did it, and how we're rolling it out one country at a time. blog.x.com/en/engineering/re…