New Constructs Calls Anthropic's IPO the 'Most Ridiculous of 2026' as HN Debates the $2 Trillion Valuation

New Constructs, an independent financial research firm, has called Anthropic's upcoming IPO the "most ridiculous IPO of 2026," arguing that the company's reported $2 trillion target valuation is an "unprecedented test of investor gullibility." The call, reported by CNBC on October 7, 2026, triggered a 216-comment Hacker News thread in which commenters split sharply over whether the critique is a rigorous analysis or a broken clock striking again.

What New Constructs actually claims

New Constructs values Anthropic at roughly $150 billion and says the company has no viable path to justifying a $2 trillion market capitalization. The firm's central argument is arithmetic: to reach its desired valuation, Anthropic would need to earn double the trailing-year profit of Nvidia, whose net income over the prior four quarters topped $190 billion.

Against that benchmark, the firm cites reported figures showing Anthropic generated $4.6 billion in revenue in 2025 while posting a net loss of $42 billion, based on a leaked copy of the company's prospectus reported by Reuters. New Constructs concluded bluntly: "We don't think Anthropic has a viable business."

The report also argues that open-source models have structurally undermined closed-model economics. "Since the arrival of open-source models, it's been clear that the closed models would struggle to generate profits," the firm wrote. It further contends the IPO's purpose is not to enrich public-market investors but to provide liquidity for Anthropic's Wall Street backers.

One unusual element of the report: New Constructs cites Anthropic's own safety messaging as a reason to avoid the stock, noting that the company's assertion that AI could pose "a catastrophic or existential risk to humanity" is itself a risk factor for investors.

The track record problem

The strongest counterargument on Hacker News was not about Anthropic's finances but about New Constructs' history. Commenter binlog compiled a list of IPOs the firm was bearish on that subsequently performed well:

New Constructs is bearish on every tech IPO. Easy to reinforce your bias by selectively looking at instances where they were right, like WeWork, but they also predicted failed IPOs for Spotify (+219%), Uber (+60%), CrowdStrike (+3136%), Cloudflare (+2306%), Palantir (+1991%), Snowflake (+51%), DoorDash (+93%), Robinhood (+187%), ARM (+422%), Instacart (+59%), Reddit (+370%), CoreWeave (+105%), SpaceX (+20%) and lots more. So whatever your stance on Anthropic this particular analysis is safe to ignore.

CNBC's own article acknowledges the mixed record. New Constructs named DoorDash its "most ridiculous" IPO of 2020 and compared it to WeWork; DoorDash's market cap subsequently grew to roughly $83 billion. The firm was also bearish on Allbirds' 2021 IPO, and Allbirds later sold its assets for an estimated $39 million. Its WeWork call, by contrast, preceded the company's pulled IPO and eventual 2023 bankruptcy filing.

Founder David Trainer addressed the misses directly in a 2021 CNBC interview, saying "crazy stuff happens" and that New Constructs does not always get it right. "I can't let that bother me," he said. "I have to stay true to what I think is right." Commenter scandox summarized the dynamic with a single line: "Monkey remains true to darts, dartboard."

The growth-rate counterargument

The second major line of pushback concerned whether the 2025 loss figures are the right basis for analysis at all. Commenter mattmaroon framed the bull case:

$2 trillion is a lot of money and certainly doesn't seem a safe bet at that price. But also, their revenue went from <$0 in 2024 to $4b in 2025 to what will likely be >15x that in 2026. Has anything ever grown that fast before? If you carry that forward even one year, and I'm not saying you certainly can, they're outselling NVIDIA.

That concern about stale data was echoed by Simon Willison, who noted that the CNBC article leans on 2025 financials while Anthropic has reportedly claimed two profitable quarters in 2026:

"The company's growing operating losses" - yeah, if you are a "research firm" basing your research on the leaked 2025 Anthropic financials and not what happened in 2026 you are on the verge of malpractice. [...] UPDATE: I found the actual New Horizons report here - it does acknowledge the 2026 numbers, unlike the linked CNBC story about that report [...] It does lean into those 2025 figures when talking about Anthropic's losses though, which is weird when Anthropic have supposedly claimed two profitable quarters in 2026.

CNBC's article does include newer figures: the New York Times reported in September 2026 that Anthropic was on pace to generate $100 billion in annualized revenue by the end of 2026, and Anthropic itself claimed in late July 2026 that its annualized revenue run rate had grown sevenfold year-over-year to $65 billion.

The unit-economics question

A separate critique focused on the structural mismatch between inference revenue and training costs. Commenter anon373839 argued that profitability depends on resolving one of two impossible options:

I don't see how Anthropic can become profitable enough to justify its investments unless they can either stop training models ("Pace the Frontier!!1"), or magically eliminate open source competition so that their volume x margin on inference is high enough to dwarf the training costs. For the IPO, I assume they're going to try to get everyone to focus on inference - which is profitable in isolation. But I don't know how they'll distract the market from the other half of the equation.

Commenter wwizo offered the opposing view, arguing that Anthropic's real product is not the model but the packaged service layer around it:

The model is engine, the harness is the car Anthropic are selling. They are a service company, selling home-grown, mass-produced, conveniently packaged intelligence at a well established price for consumer. [...] Every tech company, tech department, startup that I know is using claude code to deliver. The boutique deepseeks, qwens, glms and kimis are nice, but ain't production ready in 98% of cases.

The open-source threat, contested

Whether free open-weight models actually threaten Anthropic's pricing power was itself disputed. Commenter ViktorRay asked what advantage Anthropic retains as Chinese labs and Meta release open-weight models for free. Commenter AtNightWeCode pushed back on the premise:

I think people saying that open source is a threat underestimates how scared large corps are for lagging behind when it comes to AI. There are other risks that are bigger I think like the political risk.

Commenter chrsw argued that both the bear and bull cases are being stated with unwarranted certainty:

I don't understand how people are so certain on how this will play out. We have people basically saying companies like Anthropic offer no value because open source models are free. Then we have others saying once frontier AI labs go RSI FOOM takeoff they'll essentially become the tech industry. None of this sounds credible.

What we still don't know

The most concrete procedural objection came from commenter skybrian: "Shouldn't we wait until the S-1 is officially released before jumping to conclusions? Currently the financial data we have is based on leaks." New Constructs itself acknowledged this limitation, writing that it was "not fortunate enough to be one of the few to whom Anthropic's S-1 was selectively disclosed" and that it relied on "reports of the leaked financials." Anthropic did not respond to CNBC's request for comment.

The valuation-ceiling argument

Commenter skylake5 raised a point that does not depend on New Constructs' credibility at all: the math of a $2 trillion entry price.

10 months back 1T made sense, today even that looks expensive and they are eyeing 2T. How much do they expect to grow from here as a 10x growth from 2T is 20T and even a few years later 20T for a company is a lot. Amazon and Google ipoed with lots of room to grow.

Commenter sobiolite offered a middle path, predicting a post-IPO drawdown followed by long-term recovery, drawing an explicit parallel to Uber: New Constructs argued Uber's IPO valuation was too high, the stock initially fell, but Uber later became profitable and traded above its IPO price.

The bottom line for readers

The New Constructs report is a bearish valuation argument built on leaked 2025 financials, a $150 billion fair-value estimate, and the claim that open-source competition caps closed-model profitability. Its credibility is contested on two fronts: the firm has issued similar "most ridiculous IPO" calls on many companies that subsequently performed well, and its loss figures predate Anthropic's reported 2026 revenue trajectory toward a $100 billion annualized run rate. Anthropic's prospectus had not been made public at the time of the report, so no party in the debate had access to audited figures.