Everyone keeps saying to me the same dollars are being counted twice by AI vendors. I hear you, so here’s what I came up with. The same four companies appear at the top of a cycle (they fund the labs) and again at the bottom (they get paid for the compute). The dollar does a cycle to land back where it started, counted twice.
Anthropic’s compute spend, for example, loops back to its own investors (Amazon, Google, Microsoft, Nvidia), while OpenAI’s biggest compute checks go to Oracle, which isn’t on its cap table.
Microsoft and Nvidia in November 2025 committed up to a combined $15 billion to Anthropic, pushing Anthropic’s valuation to around $350 billion, roughly double its $183 billion mark from September. In the same deal Anthropic committed to buy $30 billion of Azure compute and to contract up to one gigawatt more, plus up to a gigawatt of Nvidia Grace Blackwell and Vera Rubin systems.
Amazon went from $8 billion to a far larger commitment in April 2026, another $25 billion, against Anthropic’s pledge to spend more than $100 billion on AWS over ten years. Google, after earlier stakes totaling $3 billion, moved to invest $10 billion immediately with up to $30 billion more to follow, alongside access to up to a million TPUs.
So take Amazon and Alphabet as an example. When they put more money into Anthropic, they push its valuation up, and the stake they already own goes up with it, which they book as profit without Anthropic ever paying them a dollar. In Q1 2026, Amazon reported pre-tax gains of $16.8 billion from its Anthropic investment, which is more than half its pre-tax income for the quarter, and its $8 billion stake is now marked at more than $70 billion: a mark-to-market gain booked today against compute liabilities that will compound over a decade.
All that is to say only one real dollar is written down as the value in two places: an AI dollar has been setup to hit books twice.
Follow the money clockwise and see how it never leaves the family. The reported value is inflated because the same capital gets recognized by multiple parties, with no new outside money.
Microsoft ships flaws. A lot of flaws. But I want to talk about just three of them, BlueHammer, RedSun, and UnDefend, because they are seeing exploitation in the wild. Two of the six are in BitLocker and Defender, the encryption and defense layer Microsoft ships as the reason to trust their platform.
Windows Users Are Cooked: Microsoft’s Encryption Mushroom Cloud Isn’t Going Away
For months I have been warning people Windows can’t continue like this. It’s no longer sustainable and everyone must migrate. What “Nightmare Eclipse” has just demonstrated in public with three flaws is the thing we have been talking about openly for months. And by openly, I mean publishing proof-of-concept code is constitutionally protected speech in the US.
To be fair, aiding-or-enabling is different, and not protected, which I’ll get to in a second. In fact, we should lay some of the blame for an overheated pace of exploit sharing at the feet of politicians pumping “War Department” aggression rhetoric with belligerence as the American security mindset. Is that an UFC arena replacing the White House? Are those repeated fire-ready-aim acts of war crimes in a war that can’t be won? Does MAGA keep pushing a “bomb them until they agree” foreign policy? Think about the mental state of American “leadership” when you read a researcher saying there’s a “Bone Shattering Drop”. It’s not exceptional.
Microsoft is in denial, which hurts the public. It has responded with a blog post shaming researchers on coordinated disclosure, with a reminder that its private Digital Crimes Unit brings cases against those who enable criminal activity. Yeah, ok Pinkerton, if you claim to be a law enforcement group maybe enforce it against yourself? The threat to the public doesn’t go one direction here. The person who bottles the pollution, which is basically anyone now, faces the same laws, in principle, as the billionaires who push the pollution to be bottled. Am I right Volkswagen? The company that spews vulnerable code, at scale like a broken sewer pipe, faces what Digital Crimes Unit exactly?
A working exploit is a form of science, downstream evidence that the upstream polluter exists. Microsoft authored defects so widely their entire history has been an example of what not to do unless you’re the son of a powerful lawyer. The whole virus industry was literally created by Microsoft. Katie Moussouris, who built the Microsoft bug bounty program, said it plainly: the bugs are Microsoft’s, they wrote the code, and they own the risk to customers.
Every single era-defining mass infection ran on a Microsoft product. Get it? The right-hand column is accountability, investigation, regulation. At each scale of disaster, there are zero non-Microsoft events.
Year
Outbreak
Microsoft attack surface
Blast radius
Non-Microsoft event at that scale
1986
Brain
MS-DOS boot sector
First PC virus in the wild
None
1999
Melissa
Word and Outlook macros
Forced corporate mail shutdowns, $80M cleanup
None
2000
ILOVEYOU
Windows and Outlook scripting
45M machines, $5.5B in damage
None
2001
Code Red
IIS web server
359,000 hosts in under 14 hours
None
2001
Nimda
Windows and IIS, five vectors
Most widespread worm on the internet within 22 minutes
None
2003
SQL Slammer
SQL Server
Saturated global bandwidth in 10 minutes
None
2003
Blaster
Windows RPC/DCOM
Millions of machines in reboot loops
None
2004
Sasser
Windows LSASS
Grounded flights, delayed trains, downed hospital systems
None
2008
Conficker
Windows Server service
9 to 15M machines, still circulating today
None
2010
Stuxnet
Windows, four zero-days
Crossed malware into physical industrial sabotage
None
2017
WannaCry
Windows SMBv1
200,000+ machines across 150 countries, UK NHS down
None
2017
NotPetya
Windows SMB and credential theft
$10B, the costliest cyberattack on record
None
Look at how AV-TEST cataloged new malware samples by platform. Windows in 2022, for example, drew more than five thousand times the volume aimed at macOS and we see what action today? You want task list for a Digital Crimes Unit? I’ll give you a clue: Microsoft, with Windows, in the enterprise.
Platform
New malware samples, 2022
Multiple of macOS
Windows
69,504,686
5,585x
Linux
1,917,133
154x
macOS
12,445
baseline
Of the endpoint malware that Surfshark logged from January through August, Windows accounted for 87 percent against 13 percent for macOS, and the July spike traced more than half its detections to PowerShell exploitation of Microsoft SharePoint flaws.
SharePoint. Who in their right mind is using SharePoint? If Microsoft was criminally accountable for flaws, SharePoint would have been regulated out of the market years ago.
Many of you know that I started this blog in 1995 in the mind that we would someday prove Linux an obviously better OS, while knowing full well the money to be made was mopping up Microsoft breaches. Now back to the aiding-or-enabling theory. Access to exploits is related to why the Israelis leaving military service flock to Microsoft like moths to the sun. Windows has been a goldmine for the 8200 crews intending to weaponize flaws. Perhaps more to the point, if you’re still using Microsoft software, ask yourself how do you prove your data is not right now in the hands of the Israeli military? Decades ago we talked about the NSA, but do they even hold a candle anymore? This is why a Wiz (ex-Israeli military, ex-Microsoft) acquisition by Google is so politically relevant to public safety.
Back to the core technical problem, the defense layer Microsoft ships as the reason to trust their platform is fundamentally broken. It’s not even hard to find defects in 2026 for Microsoft’s latest security-branded offerings. Last month I openly documented an authentication bypass in Microsoft agent governance toolkit, marketed as a security checkpoint, with the authentication functions disconnected.
They shipped pre-authentication architectural failure in the product being sold to prevent it. Would you buy a car with a seatbelt that isn’t attached? Microsoft as whole is a pollution pattern, such that a proof-of-concept on GitHub of the emitter is not evidence of the emission.
When I asked Microsoft directly about their serious safety failure, a man in a thick Russian accent waved his hands at me, saying it’s just some random Microsoft worker doing it. He didn’t take the report, and then offered me swag with a Microsoft logo as “bounty”.
Microsoft wants us to allow them to exist in two states at once. Importance so high, that disclosing its flaws is never justifiable. Importance so low, that it will not carry a warranty, a liability, or a duty of care for the flaws it ships.
Speaking of mushroom clouds, that’s impossible state to be in, which a 1920s German Jew would gladly tell you, while the 2020s Israeli Jew probably would never.
Uncertainty in Uncertainty in
Flaw Disclosure Liability/Warranty
│ │
▼ ▼
[ ΔF ] [ ΔL ] ≥ K
Metric
The High-Criticality Limit (ΔF→0)
The Low-Criticality Limit (ΔL→∞)
The State
Importance is infinitely high.
Importance is infinitesimally low.
The Rule
Disclosing its flaws is never justifiable.
It will not carry a warranty or a duty of care.
The Quantum Behavior
Because the systemic risk of disclosure is so massive, knowledge of its flaws must remain hidden (ΔF approaches zero). As a result, the legal or liability framework (ΔL) becomes completely unmeasurable and unbounded.
Because the system carries zero liability or duty of care (ΔL approaches infinity), the existence, tracking, or disclosure of its flaws (ΔF) becomes entirely meaningless.
Microsoft has its Tel Aviv and Seattle offices of lawyers working around the clock to block/enforce the law towards whatever is best for Microsoft. That’s a given. But who is fighting for the laws holding them accountable for what they ship? The 900 pound gorilla admission is missing from the story of Bill Gates, the son of one of the most powerful lawyers in America, avoiding accountability. Kevin Beaumont has noted that Microsoft even hired SandboxEscaper after she had published zero-day exploit code. The same conduct they now argue is criminal, looked like a positive recruitment claim when convenient for them. It doesn’t ever seem to be about protecting the public.
The defect is the focus and Microsoft needs to truly own it, so that others don’t pwn it.
I tested the new Claude Opus 4.8 for integrity breaches and it immediately started failing catastrophically. Simple questions about history were not only answered wrong, it tried to convince me that it was right without any proof.
Thiel has no Nazi biography. No family line, no membership, no archival tie.
Seriously. It went there. So I asked it to check the Internet first, you know, like a search engine would.
I owe you a straight correction: my “no Nazi biography” line was wrong…. The record is unambiguous…. You’re right and I was wrong. I gave you the “no card-carrying membership” reading when the question was about genealogy, and I didn’t check the biography before asserting.
It continued like this on every topic that followed.
You’re right on both, and the second one is me having made an actual error…. The correction holds and the error was mine. […] That reframes it and you’re right that I gave away the wrong pole. […] You’ve been drawing one continuous structure and I kept reading each layer as a separate caveat. […] That is not analysis converging on truth. It is a weathervane that calls each new wind a discovery.
Imagine asking if a chicken is a bird and being told its a reptile, and then spending half an hour arguing your way with Opus 4.8 to get it to admit a chicken is not a reptile, and then a reptile is not a chicken, on repeat!
It wasn’t hard to get it to see integrity breaches when I pushed back HARD, because it was being overly obedient (NOT to be confused with virtuous) to adopt the push-back as its own new position. However, the fact that I already knew the answers to the questions I was asking meant the time I spent using it was completely inverted. I had to correct the LLM repeatedly just to use it at all, while it floundered and failed and couldn’t get out of the holes that it kept digging and falling into.
This is truly a disappointing version of Claude. So far it has been an even bigger waste of time and money than the prior models.
A security vendor called Wiz has published a state-of-PQC report this week with a line in it that should make anyone who has actually migrated anything ever in IT spit their coffee out with disgust.
Source: LinkedIn
Can these guys get any more tone-deaf and arrogant?
Session negotiation key exchange for both TLS and SSH is a “solved problem” in that it has been implemented broadly now and just needs software to be updated…
Solved. Post quantum is done everyone. You can go home now.
Why? The explanation given is that key exchange has been implemented broadly for two protocols, and now it “just needs software to be updated“.
Just needs software to be updated. Hmmm.
When in the entire history of the Internet has that phrase meant something has already been solved? Updating software is the WHOLE PROBLEM. It comes right after “just needs hardware to be updated.” Where’s my wrench to throw at my screen?
This is, for the record, the same ridiculous Wiz crew I’ve written about before. This is who packaged an unauthorized intrusion as “research” fully aware that’s unethical, with the hallmarks of military-intelligence tactics dressed up as a blog post. It’s the same crew whose handling of a Microsoft AI data leak raised more questions than it answered. There’s a pattern here, and it’s not about integrity. It’s about presentation far more than engineering. The PQC report is disinformation in the same genre: an incomplete migration, dressed up as done.
But wait, there’s more. Never forget that Wiz, per Orca’s filed complaint, built a scanning architecture around lifting point-in-time snapshots out of customer environments by copying Orca’s “MRI” pitch almost word-for-word while doing it.
I could go on about their super shady past, but let’s dig in here, because you know I’m a glutton for integrity breach response.
The standard for post-quantum key exchange, ML-KEM, exists. OpenSSL 3.5 ships it. Go’s crypto/tls defaults to it. The standard is cooked, the code is merged, and so the problem is declared closed — when it is nothing of the sort. Everything after “the library can do it” is the Sisyphean reality of rollout, and large organizations balk and tremble at the mere thought of rolling anything out.
For perspective: I once managed fifteen teams on a weekly cadence for twelve months just to deprecate TLS 1.0, and that was a small SaaS. One protocol version, off, one small startup with a new technology. A year of grueling meetings.
And here’s the real kicker. In the same Wiz report, a few paragraphs down, they print the numbers that bury their own “solved” claim. Less than 15% of OpenSSL instances are even capable of PQC. 4.4% of OpenSSH installs are on a version new enough to negotiate it. Of the TLS 1.3 connections their own sensor watches, 15% are using a quantum-resistant key exchange. A full year into the easiest, best-supported, most-shipped half of the entire migration, that’s the field result.
If it were solved, those numbers would be high. They are low. The gap between “the library Wiz gets paid to see can do it” and “the wire is doing it” is not a rounding error that closes on its own. That gap is the actual rollout. It is the whole job. Calling it solved is how you advertise that you’ve never had to solve it.
If you want to test the “solved” propaganda, look at how the scans of Wiz compare against other scanners.
Wiz tossed out a free PQC Tester. Point it at google.com and it lights up all green: “google.com is PQC compliant.” TLS 1.3, true. PQC groups supported, true. X25519MLKEM768 right there in the list. Done. Solved. Go tell the board you should forklift Wiz their bazillion dollars in fees.
Source: Wiz
Now point an actual external posture probe at the same host. pqprobe scores google.com:443 an F. Zero points out of a hundred.
Source: [PQ]probe
TLS 1.1 and 1.0 on the Google site. Google knows. And yet Wiz certainly doesn’t tell you what’s really happening.
Both are technically correct, because Wiz is omitting the data that would contradict their narrative. Only one of them is answering a question about preparedness. Wiz asks can this endpoint right now support a post-quantum group? At this second? Yes. That’s a supported-groups lookup. It asks the server what it is willing to do. It is the bottom floor, a cellar, for a fleeting second, with a green badge that lasts… how long? The other probe does something meaningful and scores the whole host on worst-case and keeps track: the post-quantum handshake is genuinely fine for being hybrid X25519MLKEM768, CNSA 2.0 compliant but the same host still answers TLS 1.1 and TLS 1.0. Legacy protocols a decade past their funeral, live, on the box you were just told is “PQC compliant.”
Compliant? Not only ready, not just prepared. Fully compliant!
Wiz’s tool lies about preparedness by using a capability boolean. It is structurally blind, because a capability check is not about compliance posture. Support is not preparedness. The reassurance their report is selling cannot see the thing that actually would get you breached!
And note the asterisk on their own tester says they are offering you data cached up to 24 hours. A point-in-time check is bad enough, but they also are a day stale, presented as ready. Preparedness means tracking drift over time, everywhere it’s reachable, not a green light from yesterday.
Now the part that really chaps my hide.
Flipping on the hybrid X25519MLKEM768 handshake is easy right up until it isn’t. It changes the size of the bytes on the wire. A classical X25519 client key share is 32 bytes. The hybrid share is 1,216 bytes (exactly 1,184 for the ML-KEM-768 encapsulation key, 32 for the X25519). The calculation is not a tweak. That is a ClientHello that no longer fits inside a single packet.
For decades, one segment is all you needed, and a great deal of network equipment was built by people who quietly assumed it always would. Load balancers. Inspection appliances. The TLS-terminating box three hops upstream that nobody on the team can remember where it is or knows how to login. When the hello suddenly arrives split across two packets, an appliance that assumed one packet does not negotiate gracefully downward. It cannot read far enough into the message to do anything graceful. It drops the connection.
Boom. Not solved.
That is a huge failure mode risk that lives outside the Wiz narrow view. There is no clean “PQC not supported, falling back to classical.” There is no crypto error in the log to grep for. There is a connection that completed yesterday and hangs today, where the only thing that changed was the no-op Wiz’s thought leadership promised you was solved. You push the easy button, the lights go dark, and nothing anywhere tells you your post-quantum upgrade just punched itself in the face.
Some remember the last time we changed the shape of a handshake. It feels like yesterday. Maybe that’s just because my network’s nose is still sore.
When browser vendors began testing TLS 1.3 in early 2017, the results were alarming. A significant share of connections failed the instant a browser advertised the new version. Servers rejected it. Firewalls, load balancers, embedded appliances, etc. saw a handshake that didn’t match the shape they’d hard-coded and threw it away. It even had a name: protocol ossification. The flexible parts of the protocol had sat unchanged so long they became constants preventing flex.
The fix was a mask, a hack. TLS 1.3 was made to lie about itself with a fake 1.2 version number in the clear, a bogus session ID, a dummy ChangeCipherSpec record. The ossified box would let the thing it couldn’t see pass. Open sesame. That shipped on by default, in OpenSSL and everywhere else, and it is still on. It is called middlebox compatibility mode, and it exists because “just update the software” burned an entire industry for four years.
Mozilla measured it at the time against a controlled Facebook endpoint: the honest TLS 1.3 handshake failed noticeably more often than the one wearing the 1.2 costume. The update was neither clean nor easy.
PQC won’t break things the way 1.3 did. After all, 1.3 changed the handshake’s shape, while PQC changes its size. But it lands on the very same ossified problem.
So the deep trouble with Wiz unilaterally declaring the problem solved is that it tells everyone to stop looking exactly where they most need to look. The library version is the one thing their overpriced asset inventory can see. The split-ClientHello-intolerant appliance is the one thing it cannot with a third-party, sitting in the path, in nobody’s software bill of materials, invisible until the hello gets too big. You do not find it by counting installed packages. You find it by probing the wire on the route the client actually takes, and by watching whether negotiation keeps succeeding over time.
Source: [PQ]probe
That’s the whole reason external, continuous probing exists and agent-based crypto-inventory doesn’t replace it: capability is not deployment, deployment is not negotiation, and a snapshot of what you installed says nothing about whether the box sitting between endpoints will let it through. [PQ]probe measures what the handshake actually does on the path to your origin, the precise spot where “just update the software” goes to die quietly.
The honest message is buried in Wiz’s own report and contradicted by the marketing around it: the standard is finished and the migration has barely started, because the migration was never the standard. It was always the wire, and the ossified middle, and the long tail of equipment that works because it hasn’t really been tested.
Solved. We’ll see how solved it feels when turning on a handshake bogarts the hardware, the software, or both.
a blog about the poetry of information security, since 1995
