⏱️ Lectura: 10 min
Dan Luu, an engineer known for his data-driven essays, reviewed Ed Zitron’s predictions about Meta, Google, and Microsoft, including his November 2024 talk where he declared that Meta was “a dying product and a dying company,” and tested them against those companies’ real financial reports.
📑 En este artículo
The result, published on danluu.com, contradicts the central premise of one of the most cited AI critics right now. Between 2023 and the first half of 2026, Meta, Alphabet, and Microsoft kept growing in both revenue and profit, and that’s where the narrative of desperate companies that no longer know how to grow starts falling apart.
TL;DR
- Dan Luu compared Ed Zitron’s predictions with real revenue data from Meta, Google, and Microsoft between 2023 and 2026.
- In November 2024, Zitron claimed that Meta is a dying product and a dying company.
- Meta’s revenue went from $135 billion in 2023 to $201 billion in 2025, with double-digit growth every year.
- Meta’s operating profit grew 48% in 2024 and 20% in 2025, according to GAAP figures cited by Luu.
- Alphabet went from $307 billion in revenue in 2023 to $403 billion in 2025, with $129 billion in profit.
- Microsoft reported $305 billion in revenue in 2025, 17% more than the previous year, with $143 billion in profit.
- Zitron based Facebook’s user decline on Similarweb data, an external source that isn’t reliable for absolute figures.
- In the first half of 2026, all three companies kept growing between 10% and 30% in operating profit.
Introduction: Who Is Ed Zitron and Why His Predictions Matter
Ed Zitron became one of the most cited voices of AI skepticism through his podcast and columns, where he argues that much of the tech industry is running on an unsustainable bubble. His writing circulates frequently across technical networks and mainstream media whenever they look for a critical counterpoint to Big Tech’s optimism.
Dan Luu, the author of the analysis, discloses his own bias before getting into the details: he has no meaningful financial stake in AI companies beyond standard index funds, doesn’t work at an AI lab, and didn’t even pass a technical interview at one years ago. His stated position is simple: if something is happening right now, people who claim it’s impossible tend to be wrong. Luu had already applied this same method before: in 2022 he reviewed predictions from futurists like Ray Kurzweil and found they failed both in outcome and in reasoning, and in 2015 he wrote about how AI’s ability to displace jobs was being underestimated.
What Happened: Ed Zitron’s Predictions Under the Microscope
In the November 2024 talk that Luu uses as a case study, Zitron argues that Big Tech no longer knows how to grow and that, in its desperation to reignite growth in a dying ecosystem, the tech industry is going to shove AI into everything. He specifically names Meta as a dying company, and includes Google and Microsoft in the same diagnosis.
Luu checks that claim against the financial statements the three companies filed with the SEC during the same period. Instead of citing third-party figures or estimates, he uses officially reported GAAP revenue and operating profit, year by year, including the first half of 2026. His conclusion is direct: Zitron’s reasoning might hold up in some philosophical sense (a company can be dying without having died yet), but the available data shows no sign that Meta, Alphabet, or Microsoft are in decline.
Context and Background
The pattern Luu identifies in Zitron’s argument repeats across every prediction he reviews: it leans on minor, real but limited problems to support conclusions of a much larger scale. In Meta’s case, the cited evidence was a supposed drop in Facebook’s monthly active users, taken from Similarweb rather than the figures the company itself reports. Meta stopped publishing its official MAU in December 2023, but most available estimates show growing, not shrinking, usage, and ad revenue, which depends directly on that user base, confirms that trend.
In Google’s case, Zitron centers much of his criticism on Prabhakar Raghavan, whom he describes as a class traitor within Google’s engineering org who supposedly damaged Search’s quality. Google engineers who commented publicly on that hypothesis don’t agree that Raghavan is the sole or primary cause of Search’s quality issues. And even if that hypothesis were true, Alphabet has other business lines, like YouTube and Google Cloud, capable of sustaining growth even if Search stagnated.
The Real Numbers Behind Ed Zitron’s Predictions
The table below summarizes the GAAP revenue and operating profit for Meta, Alphabet, and Microsoft between 2023 and the first half of 2026, as compiled by Luu from each company’s public financial reports.
| Company | Period | Revenue | Operating Profit |
|---|---|---|---|
| Meta | 2023 | $135 billion (+16%) | $47 billion (+62%) |
| Meta | 2024 | $165 billion (+22%) | $69 billion (+48%) |
| Meta | 2025 | $201 billion (+22%) | $83 billion (+20%) |
| Meta | H1 2026 | $117 billion (+30%) | $42 billion (+10%) |
| Alphabet | 2023 | $307 billion (+9%) | $84 billion (+13%) |
| Alphabet | 2024 | $350 billion (+14%) | $112 billion (+33%) |
| Alphabet | 2025 | $403 billion (+15%) | $129 billion (+15%) |
| Alphabet | H1 2026 | $230 billion (+23%) | $80 billion (+30%) |
| Microsoft | 2023 | $228 billion (+12%) | $101 billion (+21%) |
| Microsoft | 2024 | $262 billion (+15%) | $118 billion (+17%) |
| Microsoft | 2025 | $305 billion (+17%) | $143 billion (+21%) |
| Microsoft | H1 2026 | $173 billion (+18%) | $79 billion (+19%) |
The pattern is consistent: none of the three companies shows a drop in revenue or profit during the period analyzed. Percentage growth even accelerates in several quarters, which is hard to square with the idea of companies that no longer know how to grow and are turning to AI as a last resort.
How to Verify the Data Yourself
Any developer can repeat this check without relying on third parties. The SEC exposes a public XBRL data API at sec.gov, where every publicly traded company in the United States publishes its figures structured by accounting concept, with the CIK as a unique identifier.
import requests
headers = {"User-Agent": "programacion-bot [email protected]"}
url = "https://data.sec.gov/api/xbrl/companyfacts/CIK0001326801.json"
response = requests.get(url, headers=headers)
data = response.json()
print(data["entityName"])
This first script just confirms you’re querying the correct entity, Meta Platforms, with CIK 0001326801. The next step is to extract a specific accounting concept, like revenue, and filter for annual reports (Form 10-K):
import requests
def revenue_por_anio(cik):
headers = {"User-Agent": "programacion-bot [email protected]"}
url = f"https://data.sec.gov/api/xbrl/companyfacts/CIK{cik}.json"
facts = requests.get(url, headers=headers).json()["facts"]["us-gaap"]
concepto = facts.get("RevenueFromContractWithCustomerExcludingAssessedTax")
for dato in concepto["units"]["USD"]:
if dato["form"] == "10-K" and dato.get("fp") == "FY":
print(dato["fy"], dato["val"])
revenue_por_anio("0001326801") # Meta Platforms
The result is a list of fiscal years with the reported annual revenue, the same kind of data Luu used to build his table. The full process, from a public claim to its verification against the primary source, can be summarized like this:
flowchart TD
A["Public claim"] --> B["Find the original source"]
B --> C["Check against primary data (10-K, 10-Q)"]
C --> D{"Does the data support the claim?"}
D -->|"Yes"| E["Claim verified"]
D -->|"No"| F["Claim refuted"]
💡 Tip: You can apply the same pair of scripts to any publicly traded company in the United States: you just need its CIK, available through the EDGAR search tool.
Impact and Analysis
The Zitron case isn’t an isolated attack on one critic. Luu reviewed several predictions beyond the November 2024 one, and the pattern repeats: an unreliable third-party figure, a personal jab at an executive, or a real but narrow problem, used as the basis for conclusions far bigger than what that data supports. It’s the same kind of methodological error Luu had found in 2022 when reviewing optimistic futurist predictions, just pointed in the opposite direction.
This doesn’t mean the AI industry is free of risk. Luu himself clarifies that his analysis is limited to checking specific claims against concrete data, not to evaluating whether capital spending on AI infrastructure is sustainable long-term, or whether circular investment deals exist between AI vendors and customers, both legitimate debates with their own evidence. The practical lesson for any developer or technical reader is simpler: before accepting a strong claim about a public company, it’s worth taking the ten minutes it takes to check its 10-K report.
📌 Note: Verifying that one specific prediction was wrong doesn’t prove the AI industry is risk-free: capital spending and circular deals between vendors are a separate debate, with their own data.
What’s Next
Meta, Alphabet, and Microsoft will keep publishing quarterly results, and each release is a new opportunity to test both optimistic and skeptical predictions about AI. The format of Luu’s analysis itself, a table of figures checked against a public claim, can be replicated by anyone with internet access and without paying for third-party data.
For readers following the AI debate, this case is a reminder that both excessive enthusiasm and blanket skepticism can, and should, be checked against the same primary sources: the reports companies are required to file with regulators.
Try it yourself: run the second script against Alphabet’s CIK (0001652044) or Microsoft’s (0000789019) and compare the result with this article’s table.
📖 Summary on Telegram: View summary
Frequently Asked Questions
Who is Ed Zitron?
He’s a public relations professional and tech commentator known for his critical stance toward the AI industry, with a podcast and columns where he argues that much of the sector runs on inflated expectations.
Who is Dan Luu?
He’s a software engineer known for data-driven technical essays, published on danluu.com, covering topics ranging from systems performance to the historical accuracy of tech predictions.
What does GAAP operating profit mean?
It’s profit calculated according to Generally Accepted Accounting Principles in the United States, the standard public companies use to report results to the SEC, without discretionary adjustments.
Why isn’t Similarweb data enough to measure active users?
Similarweb estimates traffic from panels and third-party extrapolations, useful as a rough order of magnitude but not precise enough to claim a specific decline, especially when the company itself stopped publishing its official figure.
Does this mean there are no risks in the AI industry?
No. The analysis only checks specific predictions against financial data; it doesn’t separately evaluate capital spending on infrastructure or investment deals between AI vendors.
Where can I read the full analysis?
The original article, with the complete tables and direct quotes from each prediction, is published at danluu.com/zitron.
References
- danluu.com/zitron: Dan Luu’s original analysis with the complete revenue and profit tables for Meta, Alphabet, and Microsoft.
- sec.gov: the SEC’s website where the 10-K and 10-Q reports used to verify the three companies’ figures are published.
- en.wikipedia.org/wiki/Meta_Platforms: company profile with historical context on its business.
- en.wikipedia.org/wiki/Alphabet_Inc.: context on Google’s corporate structure and its other business lines.
- danluu.com: Dan Luu’s blog with other software engineering essays and data analysis.
📱 Like this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day.
0 Comments