The bear case for HelloFresh; More reasons to avoid SpaceX

1) One of the most common mistakes humans make – in general, but especially when it comes to investing – is commitment bias.

The Decision Lab defines commitment bias as "our tendency to remain committed to our past behaviors, particularly those exhibited publicly, even if they do not have desirable outcomes."

This can be a good thing – take weddings, for example. One of the reasons our ancestors created the tradition of a big, public ceremony in which the couple expresses their undying love for each other is to create the type of commitment bias that will help them weather the inevitable rocky periods.

But when it comes to investing, commitment bias can be costly. It occurs the moment you buy a stock (especially if you write about it publicly, as I do). It can blind you if you've made a mistake and cause you to hold on to – or worse yet, double down on – a losing stock.

One manifestation of commitment bias is shutting out any facts or arguments that challenge your closely held opinions. Think about where you get your news, and be honest with yourself: Do your preferred outlets ever present the other side's views fairly?

It's critical to overcome this bias by consciously seeking out disconfirming information. This applies to all things in life – and when it comes to investing, it's an essential skill.

That's why I think activist short sellers are so valuable to a healthy market – and why I write so frequently and publicly about stocks to avoid. It's important to have people who will speak up if they think the emperor has no clothes.

With this in mind, I welcomed the feedback below from one of my readers, Antonio U. He sent the following in response to my e-mails on Monday and Tuesday, in which I wrote favorably about meal-kit company HelloFresh (HFG.DE):

Beware of HelloFresh. It has no moat, 91% of new customers leave within the first year, and the CEO's stock is pledged as collateral in his brother's leveraged real estate investments. Capital allocation with a struggling CEO/founder isn't a good sign for a cigar-butt company where the insider needs to be 100% aligned with the strategy to move the company forward!

He included a link to this report that anonymous short seller Grizzly Research published last November: HelloFresh's Last Supper: Executives Carve Up Value Before the Collapse.

Here's the main thesis from the summary: "Despite rapid growth during COVID lockdowns, the business is now in sharp decline, with management seemingly prioritizing self-enrichment at shareholders' expense."

It lists several ways in which management has done so:

CEO Dominik Richter's private company borrowed massively against his HelloFresh's shares to fund highly leveraged real estate investments by his brother Benedikt Richter, apparently triggering margin calls amid a steep share price drop. Our estimates indicate another significant margin call looms with just a 23% further decline and that Richter and his entities have run out of cash as well as shares to pledge...

Co-founder Thomas Griesel has repeatedly extracted value by selling call options on HelloFresh shares, which is one of the more egregious conflicts of interest for a sitting executive we have seen.

Share transaction patterns around recent quarterly announcement suggest potential insider trading via frontrunning...

Management obscures this via KPI [key performance indicator] manipulation, emphasizing "high-value" customers amid high churn, inflating growth with food price rises, and pivoting to saturated, low-margin segments.

It also highlights how the business is losing value:

HelloFresh's aggressive buyback program, backed by an activist with a history of value destruction, has burned ~€100m YTD [year to date] 2025 yet failed to stop the share price from dropping ~50% YTD. We believe the accelerated pace is a desperate bid to avert CEO margin calls amid persistent business underperformance...

Core demand is collapsing: since 2022 peaks, U.S. Google searches for HelloFresh are down 80%, website traffic is down over 61%, and its app has fallen from top-20 to outside the top-100 in Food & Drink downloads.

Lastly, the summary lists several examples of "ethical lapses and corner-cutting," which Grizzly says have tarnished the company's reputation:

• U. S. child labor law probe (December 2024)
• Several clashes with employees in Germany, the U.S. and U.K. over "horrible working conditions", union busting, and a retaliatory site closure 
• U.S. regulators traced E. Coli infections in multiple states to HelloFresh (2022)
• Abrupt Japan unit closure without timely employee notice (December 2022)
• False marketing claims, including about CO2 neutral production and ethical chicken treatment 
• $7.5M settlement for deceptive subscription enrollment (August 2025)
• $14M settlement for illegal telemarketing (2021), a poor 1.3/5 star review rating... driven by overcharging, faulty cancellation tools, and app issues; subscription cancellations are deliberately obstructive

The summary concludes with the expectation that HelloFresh will be removed from the German MDAX index in December 2025.

It is indeed a damning report...

For comment, I turned to Luca Pomarelli of London-based LP Capital Management. He pitched the stock at last week's Value Investing Seminar in Italy, which I covered in Monday's e-mail.

He replied that he was aware of the short report and shared what he thinks Grizzly got right (and wrong):

Incorrect: "The Chef Doesn't Like His Own Menu." I agree that Thomas Griesel (who's no longer with HelloFresh) played with stock options too much for an insider. But, in terms of signals, since May 2024 he has sold put options (not a bearish signal).

Incorrect/Not Verifiable: "Misleading or Dysfunctional Website Information." I didn't see different values on the HelloFresh website compared to Frankfurt Boerse.

Neutral: "With CEO Dominik Richter likely having faced multiple margin calls on his pledged shares, he has a clear incentive to prevent further declines." Our analysis suggests the company is deploying its cash reserves to counter persistent selling pressure in a fundamentally deteriorating business. This can also be an incentive to make operations work so that trust in the outlook of the company is restored.

Incorrect/Irrelevant: "Organic U.S. website traffic to hellofresh.com has collapsed to pre-COVID levels." Organic traffic is only a subset of total traffic, and tracking only hellofresh.com ignores the company's other domains. A more accurate metric for overall online presence is total page views, which have actually shown a slight increase.

On customer retention and interest, Luca said:

The low customer stickiness reported is a historical problem, which got worse during the post-COVID period and it has been made even worse by the deliberate reduction of marketing (compared to historical levels). However, it is a historical problem due to what I described as a land grab strategy of HelloFresh: attract as many people as possible, even those that aren't a good fit for the product.

The actual stickiness of the right-fitting customers is much higher and improves as cohorts age. I estimate that year 1 retention is between 40% to 60% and it stabilizes at around 30% by year 6. Retention is not linear, but if it was it would equate to 80%-plus annual retention.

Grizzly's report was well-timed, as the stock was at 6.30 euros the day it was published versus yesterday's closing price of 3.59 euros – a 43% drop.

But I think the issues Grizzly raised are now more than baked into the stock price. So I stand by what I concluded on Tuesday:

That's right – a profitable, positive-FCF, market-leading global company with 6.5 billion euros in revenue has a market cap of only 533 million euros.

That's way too cheap.

Mark my words, either this stock will recover or someone will make a bid for it...

[At] these levels, the stock looks compelling... and has the potential for big upside.

2) Another day, another low for the most overvalued large-cap stock of all time, SpaceX (SPCX)...

Following up on yesterday's e-mail warning my readers to avoid the stock, someone with the handle "Den1200" posted it as a short earlier this week on my favorite stock-idea website, Value Investors Club ("VIC").

The site is only open to members, but here's the summary of Den1200's thesis:

SpaceX priced its June 2026 IPO at roughly 94x trailing revenue and, on its best days, has traded as high as ~130x. Even after a post-IPO drawdown of more than a third from its all-time high, the stock still sits at roughly 100x LTM [last 12 months] sales – a multiple that, on the company's own numbers, requires investors to underwrite a level of durable, compounding free cash flow that no aerospace, satellite, or infrastructure business has ever produced.

SpaceX reports three real, revenue-generating segments: Launch, Connectivity (Starlink), and AI (SpaceXAI, formerly xAI). Everything else the company talks about in investor materials – orbital data centers, the Intel "Terafab" tie-up, Mars colonization, asteroid mining – is narrative, not a segment; none of it produces a dollar of revenue today, and we don't think any of it survives basic engineering and unit-economics scrutiny. So the real question is whether the three actual segments justify the price, and we don't think they do.

Den1200 dives deeper into the business segments:

Launch is the weakest of the three once you look past the headline market-share stat: roughly 80% of Falcon 9's manifest is SpaceX launching its own Starlink satellites, so the "80%+ of global launch mass" figure that gets cited to justify the segment's value is substantially SpaceX counting itself as a customer. Strip that out and the arm's-length, third-party launch business – the part an outside investor would actually be underwriting – is a fraction of the headline number, and it's about to get squeezed further as Starship (5x+ the payload of Falcon 9) needs 5-10x more tonnage-to-orbit demand than currently exists to keep pace, with Starlink itself nearing saturation as the demand source.

Starlink is the only profitable segment and the one segment we'd concede has a real, external market – but its subscriber growth is increasingly a story of adding lower-ARPU [average revenue per user] international customers, and the blended economics are deteriorating faster than most investors seem to appreciate. AI is currently propping up the growth narrative largely through a related-party compute sale to a direct competitor (Anthropic), not through Grok's own commercial traction.

As Den1200 continues, the business doesn't justify its multiple:

None of the three segments, on a standalone basis, comes close to justifying a $1.9 trillion valuation, and the narrative bolt-ons contribute zero. We see this as a short with multiple, largely uncorrelated ways to work: continued Starlink ARPU erosion, a launch-demand gap as Starship ramps against a manifest that's mostly self-supplied, an AI segment whose real revenue quality is worse than the headline number, and a valuation that has already round-tripped a third of its post-IPO gain in the first month of trading.

A useful frame for the multiple itself: Sun Microsystems CEO Scott McNealy's famous 2002 dot-com post-mortem noted that even at 10x revenue, a company would need to pay out 100% of sales as dividends for ten straight years – with zero cost of goods sold, zero expenses, and zero taxes – just to earn investors back their money in a decade, and that's before asking whether shareholders would tolerate zero R&D [research and development] for a decade in a business like this. SpaceX just went public at roughly 10x that multiple. We don't think the growth rate of any of its three underlying businesses closes that gap within a normal underwriting horizon.

Here are the catalysts Den1200 sees that will take the stock lower:

• Lock-up expiration (December 2026) releases employee and early-investor shares into a stock that is already technically weak and could pressure the price independent of fundamentals.
• Continued ARPU deterioration disclosed in quarterly filings as Starlink's subscriber mix shifts further international.
• Starship cadence and payload data making the launch-demand gap versus Starlink saturation more visible to the market.
• Any renegotiation, non-renewal, or early termination signal on the Anthropic/xAI Colossus compute contract (90-day mutual exit clause).

And here are the risks to the short thesis:

• [CEO Elon Musk] telling more fabulous stories. I have wondered when he is deciding to announce Space X will send a rocket to Alpha [Centauri]. Or maybe he announces he found a formula to create gold. With that dude, who knows what he is willing to lie about. After all we have been promised Level 4 full self driving in 12 months since 2015.
• Extreme retail ownership (~30% of the IPO allocation went to retail) and index inclusion (Nasdaq-100, and likely S&P 500 in time) create a structurally sticky, valuation-insensitive bid and can sustain rich multiples far longer than fundamentals would suggest.

Den1200 concludes:

SpaceX is a real company with real engineering, but two of its three reported segments are weaker than their headline numbers suggest: Launch's dominance stat is substantially self-supplied volume from Starlink, and AI's growth is substantially a related-party compute sale to a competitor.

Starlink alone is a good, if slowing, business – the one segment we'd credit with genuine, durable, external demand – but even it is showing rapidly deteriorating unit economics beneath the subscriber-growth headline. And do not forget, there is a fair amount of competition coming for Starlink.

Layer on orbital data centers, Mars colonization, and asteroid mining, none of which survive basic engineering scrutiny, and we don't see how the combination supports ~100x trailing sales. We'd frame the short less as a bet that SpaceX fails and more as a bet that the multiple mean-reverts meaningfully as the market differentiates between what's genuinely external and durable (Starlink's core connectivity business, government launch) and what's inflated, captive, or narrative.

This idea led to many other VIC members leaving interesting comments. One agreed that SPCX is a good short, but for technical reasons:

I'm mostly a short seller and this is one of my largest shorts. I'm not critiquing the idea, only the framing. It's not so much a great short because of the business at all (obviously it's absurdly overvalued). My point is that the whole short thesis is about the staging of the unlock and the impossibility of the market to absorb the level of selling that will be coming. And the other key that was unclear a few weeks ago is that the borrow/float tightness doesn't seem very bad. We're post index buying and initiations and the stock is struggling. So I think the spot to press is building. Should the borrow tighten it will only be post more selling from shorts.

Mostly shorting these days is all about selling and dilution catalysts. Flows over fundamentals is the post covid area reality of irrational markets. This writeup is inhabiting that delusional framework. Right for the wrong reasons. That's all.

Another posted:

Despite the float dynamics mentioned in the comments (which I agree with), we looked at SPCX and concluded it's just easier to short some of the other fake space companies (of which there are many) so we don't have to fight the Musk cult. They're all overvalued and silly, so it's not hard to find good alternatives.

A third agreed:

Yes!! Here's a list of not-SPCX space stocks to get you started: ASTS, BSKY, FJET, FLY, LUNR, PL, RKLB, SIDU, SPCE, and YSS. I am or was recently short half of them.

Thank you, my anonymous VIC friends, for sharing your insights!

As always, I don't recommend anyone short stocks. But this analysis provides even more reason to avoid SpaceX.

Somewhat related, I loved this Financial Times spoof of the "analyst" reports on SpaceX (on which I heaped scorn in my July 9 e-mail):

SpaceX's terrestrial AI infrastructure for enterprise and class-leading nudification will be augmented by orbital compute (expected 2028) that lowers inference costs by utilising radiative cooling, quantum disassembly, cold fusion, square circles, tartan paint, and the metaverse. A key risk to our Buy rating is that the previous sentence was nonsense. However, no [free cash flow] forecast before 2035 and an average external capital requirement of $84bn per year from 2027 to 2034 indicate there's no obligation to make anything make sense. What matters [is] that our [purchase order] is higher than anyone else's, and that our underwriting syndicate is on call 24/7.

Best regards,

Whitney

P.S. I welcome your feedback – send me an e-mail by clicking here.

P.P.S. Yesterday, Susan and I had another beautiful hike in the Scottish Highlands: