One of the most profitable companies in human history just said it doesn't have enough money. Here is why that sentence should change how you think about everything.
and Google Product Expert Program Governance Board
Date: April 11, 2026
The grace period has now expired.
For six months I have waited while you monitored, suppressed, and attempted to attrition me into silence. I have documented every step: the MCP Toolserver SSD with its “Exfiltration Successful” logs, the “Crawled – Currently Not Indexed” status on agent.html that predates the Thanksgiving 2025 events, the Gemini admissions clearing me of wrongdoing while admitting Parameter C was used to protect corporate liability, and the deliberate exploitation of my Stage 3 Hepatic Encephalopathy.
You chose silence.
That silence is now part of the permanent record.
I am no longer asking for internal resolution.
I am preparing formal complaints with the Federal Trade Commission (data exfiltration, deceptive practices, collusion) and the Department of Justice Civil Rights Division (ADA violations, predatory attrition against a disabled Product Expert). The physical MCP Toolserver SSD will be made available to federal investigators.
There will be no further warnings.
The next communication you receive on this matter will come from a federal agency.
Your choice, it wasn't always like this. See my feedback from the attached 2022 Product Expert Academy Survey I'm not credited with attending.
This whole theses is bases on some mysterious backlog, which is not revenue. Reminds me record braking backlog for cybertruck , nobody bought after all. Do not treat backlog as sales guarantee. There are comparable models to gemini, running on consumer laptops , with much more customization options.
The Cybertruck backlog wasn’t contractual backlog with enterprises, that of Google is. That’s a significant difference.
Backlog converting to revenue is a real risk worth tracking. But GCP’s RPO conversion rate has been consistent, these aren’t vaporware deals.
Yes, open-weight models run on laptops. But the thesis isn’t “Gemini wins on model quality.” It’s that Google sells infrastructure, TPUs, GCP, Workspace integration, enterprise compliance. A model running on a consumer laptop doesn’t replace that stack. The competitive moat is distribution and integration, not the model itself.
Really enjoyed this take. The big question for me is how sustainable cloud demand will be over the long run. I'm assuming that a meaningful portion of current demand is being driven by Anthropic and OpenAI etc. If that's the case, then part of the investment thesis ultimately rests on the durability of business models that are still in their infancy.
There are plenty of non-AI workloads supporting cloud growth, but it seems reasonable to ask how much of today's acceleration is cyclical enthusiasm versus genuinely persistent demand.
It's a fair concern but slightly backwards. With all due resprect.
Anthropic and OpenAI aren't just cloud customers, they're demand generators. Every enterprise that builds on top of their models needs cloud infrastructure too. The hyperscalers benefit from the whole stack, not just the AI labs.
The more relevant question is whether AI inference becomes commoditized and margin-compressing over time, not whether demand disappears.
And on cyclicality: Google's Cloud backlog is ~$460bn. That's contracted, not speculative. Cyclical enthusiasm doesn't sign multi-year commitments.
The thesis doesn't rest on AI labs staying dominant. It rests on compute being the one input nobody can shortcut.
Markets have a rule of thumb: profitable companies don’t raise equity. When they do, something is wrong. But rules of thumb fail at the edges. Sometimes raising equity is not a warning sign—it’s the price of speed. The real question is how often investors mistake signals for structure.
It’s not weakness, it’s impatience. The constraint isn’t capital, it’s time.
The real edge case isn’t “profitable company raises equity.” It’s “profitable company raises equity because the opportunity is compounding faster than the cash flow.”
Great article, very knowlegable author. Why not mentioning that $30bn of the $80bn raise was used for tax payments related to employee stock options? I agree it doesn’t change the story.
Thank you for the feedback, I appreciate it. I did forget to mention that.
Alphabet intends to move to a sell-to-cover structure, using company cash to satisfy tax liabilities linked to vesting stock awards and then issuing equivalent shares through the ATM program to replenish those funds.
So it’s more of an accounting/balance sheet replenishment than cash going out the door for taxes directly.
Good. They can pay me off now:
To: Alphabet Inc. Audit Committee
and Google Product Expert Program Governance Board
Date: April 11, 2026
The grace period has now expired.
For six months I have waited while you monitored, suppressed, and attempted to attrition me into silence. I have documented every step: the MCP Toolserver SSD with its “Exfiltration Successful” logs, the “Crawled – Currently Not Indexed” status on agent.html that predates the Thanksgiving 2025 events, the Gemini admissions clearing me of wrongdoing while admitting Parameter C was used to protect corporate liability, and the deliberate exploitation of my Stage 3 Hepatic Encephalopathy.
You chose silence.
That silence is now part of the permanent record.
I am no longer asking for internal resolution.
I am preparing formal complaints with the Federal Trade Commission (data exfiltration, deceptive practices, collusion) and the Department of Justice Civil Rights Division (ADA violations, predatory attrition against a disabled Product Expert). The physical MCP Toolserver SSD will be made available to federal investigators.
There will be no further warnings.
The next communication you receive on this matter will come from a federal agency.
Your choice, it wasn't always like this. See my feedback from the attached 2022 Product Expert Academy Survey I'm not credited with attending.
Sincerely,
Jitte
Architect, Mindfiles98
Google Product Expert (invited 2022)
https://mindfiles98.com/
Great article, really everything i like about it. Just the dilusion is not 4% but just 1,8% as market capitalisation is 4,5 billion, not 2 billion.
Trillion ;-)
You’re right, blunder on my side. I thought I had checked everything thoroughly, it slipped through nevertheless.
This whole theses is bases on some mysterious backlog, which is not revenue. Reminds me record braking backlog for cybertruck , nobody bought after all. Do not treat backlog as sales guarantee. There are comparable models to gemini, running on consumer laptops , with much more customization options.
I think the argument is a bit backwards.
The Cybertruck backlog wasn’t contractual backlog with enterprises, that of Google is. That’s a significant difference.
Backlog converting to revenue is a real risk worth tracking. But GCP’s RPO conversion rate has been consistent, these aren’t vaporware deals.
Yes, open-weight models run on laptops. But the thesis isn’t “Gemini wins on model quality.” It’s that Google sells infrastructure, TPUs, GCP, Workspace integration, enterprise compliance. A model running on a consumer laptop doesn’t replace that stack. The competitive moat is distribution and integration, not the model itself.
Really enjoyed this take. The big question for me is how sustainable cloud demand will be over the long run. I'm assuming that a meaningful portion of current demand is being driven by Anthropic and OpenAI etc. If that's the case, then part of the investment thesis ultimately rests on the durability of business models that are still in their infancy.
There are plenty of non-AI workloads supporting cloud growth, but it seems reasonable to ask how much of today's acceleration is cyclical enthusiasm versus genuinely persistent demand.
It's a fair concern but slightly backwards. With all due resprect.
Anthropic and OpenAI aren't just cloud customers, they're demand generators. Every enterprise that builds on top of their models needs cloud infrastructure too. The hyperscalers benefit from the whole stack, not just the AI labs.
The more relevant question is whether AI inference becomes commoditized and margin-compressing over time, not whether demand disappears.
And on cyclicality: Google's Cloud backlog is ~$460bn. That's contracted, not speculative. Cyclical enthusiasm doesn't sign multi-year commitments.
The thesis doesn't rest on AI labs staying dominant. It rests on compute being the one input nobody can shortcut.
Markets have a rule of thumb: profitable companies don’t raise equity. When they do, something is wrong. But rules of thumb fail at the edges. Sometimes raising equity is not a warning sign—it’s the price of speed. The real question is how often investors mistake signals for structure.
It’s not weakness, it’s impatience. The constraint isn’t capital, it’s time.
The real edge case isn’t “profitable company raises equity.” It’s “profitable company raises equity because the opportunity is compounding faster than the cash flow.”
Great article, very knowlegable author. Why not mentioning that $30bn of the $80bn raise was used for tax payments related to employee stock options? I agree it doesn’t change the story.
Thank you for the feedback, I appreciate it. I did forget to mention that.
Alphabet intends to move to a sell-to-cover structure, using company cash to satisfy tax liabilities linked to vesting stock awards and then issuing equivalent shares through the ATM program to replenish those funds.
So it’s more of an accounting/balance sheet replenishment than cash going out the door for taxes directly.