From: Simba@techonion.org
To: Sam@openai.com
Subject: THE AGI DELUSION!
Dear Sam Altman,
Congratulations are in order for successfully convincing the entire gullible world that OpenAI is merely months away from creating an artificial god (AGI) while hemorrhaging a modest $5 billion this year. As long-time observers of Silicon Valley’s reality distortion fields, we must say yours has achieved a luminosity that would make Steve Jobs reach for his sunglasses from beyond the grave.
That Awkward Moment When Your Sugar Daddy Starts Dating Other AI Companies
Remember January 2025? That magical month when Microsoft – after injecting $13+ billion into OpenAI – casually announced it was time to see other AI models? We couldn’t help but notice the subtle shift from “exclusive cloud provider” to “right of first refusal,” which in relationship terms is like going from “married” to “I’ll call you if my date cancels.”1
What’s particularly delightful is watching your PR team reframe this development as “evolving the partnership,” which is the corporate equivalent of claiming “we’re still good friends” after finding your spouse’s profile on Tinder. Microsoft is developing its own AI models while you are busy promising AGI by next month’s Tuesday – a classic case of hedging bets while nodding enthusiastically in meetings on Microsoft Teams.
One can’t help but wonder if this shift happened after Microsoft executives googled or prompted co-pilot to do deep research on “sunk cost fallacy” and realized they had built their entire AI strategy around a company whose board once fired its CEO for 72 thrilling hours. The rollercoaster that was November 2023 certainly gave new meaning to the phrase “the best bromance in tech.”2 Nothing says true love like corporate governance chaos and threatened withdrawal of billions in funding.
The Secret AGI Timeline Calculator: Add 5 Years and Multiply by Venture Capital Needs
Speaking of things that don’t exist yet, let’s discuss your refreshingly flexible approach to AGI timelines. We have developed a formula for decoding Silicon Valley AGI predictions: Take the public estimate of when they expect AGI, add five years, then multiply by the company’s immediate funding needs.
You have masterfully avoided being pinned down to specific years while still managing to create FOMO by claiming OpenAI “knows how to build AGI” – it’s just a matter of execution!3 This statement has the beautiful quality of being both impossibly grandiose and completely unfalsifiable.
Meanwhile, your competition is getting specific. Google DeepMind’s Demis Hassabis says AGI is 5-10 years away.4 Various forecasters and AI experts are betting on 2027, 2030, or 2040. It’s almost as if everyone in AI has adopted the doomsday cult approach to predictions: keep pushing the date back when the world doesn’t end on schedule.
The true stroke of genius was realizing that if you named your models numerically (GPT-1, 2, 3, 4), eventually you would reach GPT-5, at which point people might reasonably ask, “Is this AGI yet?” So instead, we got GPT-4o, GPT-4 Turbo, GPT-4.5, and the utterly baffling o1 naming scheme. It’s like watching a tech company frantically take detours to avoid reaching its own stated destination.
The “Sorry About Your Job But Have You Considered Learning to Code?” PR Strategy
Your candid admission that “jobs are definitely going to go away, full stop” was a refreshing departure from the tech industry’s usual “no one will be replaced” platitudes – unlike the retired Bill Gates who is taking every opportunity to remind us of the impending doom.5 What made it truly special was immediately following this with the assurance that “better jobs” would be created – presumably ones that involve supervising the AI that took your original job.
This PR approach has all the empathy of telling someone whose house just burned down that they should be excited about the opportunity to go on Pinterest and start pinning to upgrade their interior design aesthetics. Goldman Sachs estimates 300 million jobs could be disrupted by AI. I’m sure all those people are thrilled about the prospect of “better jobs” that they’re not qualified for and that may not actually exist.
The true masterstroke of this messaging is how it manages to simultaneously alienate both the workers who fear displacement AND the businesses you’re trying to sell AI to. It’s like creating a product slogan that says, “Our software: It’ll fire your employees AND eventually make you obsolete too!” Marvel at how this messaging creates fear-based adoption while building a reservoir of resentment that will absolutely never backfire!
The Eastern Front: When Your Competition Speaks Mandarin and Charges 95% Less
While you’ve been busy navigating Microsoft relationship counseling and AGI prophecies, something fascinating has been happening in China. DeepSeek released an R1 model that outperforms some of OpenAI’s offerings at approximately 3% of the cost. That’s not a typo – they’re charging $2.19 per million tokens versus your $60.6
Chinese AI firms have cut their development gap from 6-9 months behind to just 3 months in early 2025.7 But the truly interesting part isn’t just the performance – it’s the approach. While OpenAI jealously guards its models behind proprietary walls, the Chinese AI ecosystem is embracing open-source frameworks, creating what analysts are calling an “Android moment” for AI.8
Your exclusive, expensive, closed-source approach is starting to look like the AI equivalent of selling $1,500 smartphones in a market suddenly flooded with $50 alternatives that do 95% of the same things. The geopolitical tariffs and export controls that were supposed to maintain Western AI advantage have instead created a parallel ecosystem that’s now threatening to outcompete you on price, openness, and soon, performance.
Even more delicious is that OpenAI seems to have anticipated this, evidenced by your latest forecast to triple revenue to $12.7 billion in 2025. One has to admire the optimism of projecting revenue growth at the exact moment your competitive advantage is evaporating and your prices are becoming indefensible.
The Stargate to Nowhere: A $500 Billion Infrastructure Play That’s Definitely Not Desperation
January 2025 also brought us The Stargate Project, your brilliant strategy to build $500 billion in AI infrastructure with SoftBank, Oracle, and others – conspicuously not funded by Microsoft.9 Nothing says “our partnership is stronger than ever” like running off to build half-trillion-dollar data centers with someone else.
The timing couldn’t be more perfect – right as your Microsoft “bromance” shows signs of “fraying,” you’ve found new friends with deep pockets and an even deeper willingness to believe in AGI timelines. SoftBank’s involvement is particularly reassuring, given their impeccable track record with WeWork, another company that promised to revolutionize a fundamental aspect of human existence.
But what caught my attention was the incredible bargaining power this gives you when negotiating with Microsoft. According to your recent investor update, you plan to cut Microsoft’s revenue share by at least 50% by decade’s end.10 Nothing motivates a partner like publicly announcing you’ll be giving them half as much money – relationship experts call this “the ultimatum approach to negotiation.”
In Conclusion: The Reverse Turing Test
As you navigate these complex waters, Sam, I’d like to propose a thought experiment: What if the true test of artificial general intelligence isn’t whether a machine can convince humans it’s intelligent, but whether a CEO can convince investors his AI is nearly sentient while actually being nowhere close?
By that measure, you’ve already achieved AGI. Your ability to maintain a $80+ billion valuation while losing billions, predicting technological singularities that perpetually remain 5 – 10 years away, alienating your biggest investor (and Overlord), and watching cheaper competitors eat your lunch is itself a form of intelligence beyond normal human capacities.
Perhaps the true innovation of OpenAI isn’t technological but financial: you’ve discovered that claiming to be building AGI is far more profitable than actually building it, at least in the short term. The question is whether you can keep this delicate balance – between hype and reality, between Microsoft dependency and independence, between proprietary advantage and open-source competition – before the whole elaborate construction collapses under the weight of its contradictions.
In the meantime, I eagerly await GPT-4.75, GPT-4.75 Supreme, GPT-4.75 Turbo Max Plus, or whatever name you choose to avoid reaching the numerically dangerous territory of GPT-5, where the AGI promises would need to be fulfilled or finally abandoned.
With a mixture of awe and bewilderment,
[TechOnion]
P.S. Have thoughts on Sam Altman’s AGI claims or OpenAI’s fraying Microsoft romance? Think Chinese AI will eat OpenAI’s lunch? Comment below with your predictions about when AGI will actually arrive, and whether OpenAI will still exist by then.
Support independent tech satire! Unlike OpenAI, we don't have $13 billion in Microsoft funding or a $500 billion Stargate Project – just a burning desire to peel back the layers of tech absurdity. Donate any amount to TechOnion, and we promise to use it for data center costs and not for building questionably sentient AI that will predict the exact moment your job becomes obsolete.
References
- https://www.businesstoday.in/technology/news/story/microsoft-is-developing-its-own-ai-models-to-compete-with-openai-report-467362-2025-03-10 ↩︎
- https://www.nytimes.com/2024/10/17/technology/microsoft-openai-partnership-deal.html ↩︎
- https://fortune.com/2025/04/15/ai-timelines-agi-safety/ ↩︎
- https://www.cognitivetoday.com/2025/04/artificial-general-intelligence-timeline-agi/ ↩︎
- https://www.businessinsider.com/chatgpt-sam-altman-jobs-replaced-ai-openai-2023-7 ↩︎
- https://time.com/7210296/chinese-ai-company-deepseek-stuns-american-ai-industry/ ↩︎
- https://cointelegraph.com/news/openai-expects-revenue-triple-competitors-catching-up ↩︎
- https://dig.watch/updates/chinas-ai-industry-is-transforming-with-open-source-models-challenging-the-openai-proprietary-approach ↩︎
- https://www.computerweekly.com/news/366621597/Microsofts-fraying-relationship-with-OpenAI-blamed-for-datacentre-expansion-plan-rollback ↩︎
- https://www.reuters.com/business/openai-plans-slash-revenue-share-microsoft-information-reports-2025-05-07/ ↩︎
