We are short SoundHound AI, Inc. $SOUN. Our full report is now available on our website, culperresearch.com
2) $SOUN is a flailing AI wannabe trading at 22x revenues that has misled investors regarding both its claims to possess revolutionary AI technology and a massive backlog of $336M in contracts averaging 6.5 years in length. We think both $SOUN's tech and its backlog are bogus.
3) $SOUN claims that it sees "immediate scaling" at restaurants, but the only large restaurant group $SOUN names is White Castle, signed in 2020. Former employees told us that White Castle was merely a trial, and remains contained to just 3 of 300+ locations, 3 years later.
4) Similarly, in early SPAC materials, $SOUN named Sonic Drive-In as a customer, yet has removed Sonic's logos from all of its materials and has not mentioned Sonic in any venues since that time. We think Sonic likely left SoundHound behind, as $SOUN's tech simply does not work.
5) Customers at the sole named White Castle location utilizing $SOUN's AI ordering say they "hate it", it's "always messed up", "did not get my order right", "is a pain", and gets orders "all wrong." Former $SOUN employees estimated the AI messed up 30% to 50% of orders.
6) $SOUN touts its offering as "fully-automated", but we discovered $SOUN uses undisclosed fully-staffed call centers. As recently as May, $SOUN posted jobs seeking a "Merchant Operations Manager" who would explicitly contract with and manage call centers.
7) $SOUN claims it's AI has reduced bad responses "to a negligible amount", but even $SOUN's chat app is useless. The app failed in even basic asks, telling us Joe Biden was the first POTUS, and naming CLOSED restaurants when we asked for food options open now.
8) $SOUN misleads on its supposed $336M backlog consisting of contracts averaging 6.5 years in length. $SOUN previously called its backlog "understated", but now forecasts just $43-50M in 2023 revenues vs. its original $110M guide, a 58% whiff.
9) $SOUN's claims literally don't add up: $SOUN says its longest deal is 9+ years, and that avg. length expanded from 4 to 6.5 years, implying $SOUN has been signing 12 to 20-year contracts. Both can't be true. A former employee corroborated 10-year contracts out of question.
10) $SOUN claims backlog "automatically rolls into the P&L" yet our convos with 3 former $SOUN employees contradicted these claims. Per formers, backlog included "potential revenue based on potential engagements" which "might be vague", and "may not actually happen."
11) $SOUN's auditor is the problematic Armanino LLP, now infamous for auditing FTX US, a bevy of microcap blow-ups, and providing aattestation services for cryptos like Nexo, later fined $45M by the SEC. Coincidentally, both Armanino's 2021 audit inspections found deficiencies.
12) Our read of $SOUN's 8-K accompanying its new $400M shelf implies $SOUN burned $16.3M in Q2, up from $14.5M in Q1, despite the Company's claims to ramp profitability thru 2023. Sell-side models we reviewed bought into management's bait-and-switch; steep revisions are coming.
13/13) $SOUN's CRO left in February, and $SOUN has yet to hire a replacement. Yet in March 2023, multiple insiders instead set up stock sale plans, and have collectively sold $11.5M in stock, most of it since June. We're short and think shares are headed lower.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
1) Last night, Rumble $RUM reported Q1 2023 results which reaffirm our thesis. We've posted an update on our website culperresearch.com
2) We previously opined that $RUM was massively overstating MAUs: $RUM claimed 80M, we estimated 38M to 48M. Almost magically, after being called out, $RUM now says users fell a massive 40% QoQ from 80M to just 48M.
3) $RUM claims that its user base disintegrated due to the impact of midterm elections. This is an asinine, nonsensical explanation: $RUM had already reported 71M MAUs in Q3 2022 (i.e., pre-midterms). We view the figure as a de facto restatement.
1) We are short Rumble Inc. ($RUM). Our full report is now available on our website. culperresearch.com
2) With little revenue generation ability, $RUM's promotion is premised on its claim to have grown to 80M MAUs. But based on numerous sources of independent website traffic and app data, we think this 80M number is a mirage, and $RUM overstates its user base by 66% to 108%.
3) $RUM reported supposedly record-breaking growth in the run-up to its de-SPAC transaction, yet leading third-party traffic and app download data suggest $RUM's actual growth has remained relatively stagnant.
We are short Xometry, Inc. ($XMTR). Our full report is now available on our website culperresearch.com
2) $XMTR is a 2021 IPO which claims to generate recurring revenues, 6.1x LTV to CAC ratios, and operating leverage on a $2T TAM. We think these claims are all bogus; XMTR is an unviable business which has unraveled while management has sold millions in stock.
3) $XMTR is a middleman for cost-conscious engineers shopping prototypes. $XMTR's "% of revenues from existing Accounts" metric is contrived and misleading; Accounts are employers, not Buyers (engineers), while XMTR continually acquires and re-acquires new engineer Buyers.
1) We are short Treace Medical Concepts Inc $TMCI. Our full report is now available at culperresearch.com
2) $TMCI claims to be a cutting-edge medtech company. We think Treace's core innovation is rebranding a decades old procedure (Lapidus fusion, as "Lapiplasty"), then implementing a dubious DTC marketing program, while surgeons employ problematic billing practices.
3) $TMCI's Lapiplasty kits cost ~$5,700, a massive mark-up that makes kits unprofitable for many surgeons. As such, we believe adoption has been accompanied by billing under "complexity adjustments" and other aggressive coding techniques that inflate reimbursement to ~$12,000.
1) We are short Veru Inc $VERU, and have published a new, updated report on our website culperresearch.com
2) On July 6, $VERU published additional data from its Phase III Study of sabizabulin in hospitalized COVID-19 patients. This was interpreted as bullish by investors and ever-obsequious sell-side analysts, yet we think otherwise...
3) $VERU claimed its placebo group and treatment groups contained "similar baseline characteristics", yet instead we find the placebo group contained a fat left tail of markedly sicker patients. A picture is worth 1,000 words…
We are short Piedmont Lithium $PLL. Our full report is now available at culperresearch.com
2) $PLL touts its "centerpiece" Carolina mine as "the best lithium asset on the planet"; we find this laughable given the mine's astronomical 11.6x strip ratio, measly 1.08% grade, and 11-year life. An expert we spoke with called $PLL's lowest-cost claims "bullshit."
3) $PLL's Carolina mine needs both a state mining permit and local rezoning. We think local commissioners deny $PLL's bid for rezoning, amid both PLL's carpetbagger behavior and the environmental and economic concerns of the mine. Without rezoning, $PLL's Carolina plans are dead.