TechCrunch : Meta bets big on AI with custom chips — and a supercomputer

Meta bets big on AI with custom chips — and a supercomputer

At a virtual event this morning, Meta lifted the curtains on its efforts to develop in-house infrastructure for AI workloads, including generative AI like the type that underpins its recently launched ad design and creation tools.

It was an attempt at a projection of strength from Meta, which historically has been slow to adopt AI-friendly hardware systems — hobbling its ability to keep pace with rivals such as Google and Microsoft.

“Building our own [hardware] capabilities gives us control at every layer of the stack, from datacenter design to training frameworks,” Alexis Bjorlin, VP of Infrastructure at Meta, told TechCrunch. “This level of vertical integration is needed to push the boundaries of AI research at scale.”

Over the past decade or so, Meta has spent billions of dollars recruiting top data scientists and building new kinds of AI, including AI that now powers the discovery engines, moderation filters and ad recommenders found throughout its apps and services. But the company has struggled to turn many of its more ambitious AI research innovations into products, particularly on the generative AI front.

Until 2022, Meta largely ran its AI workloads using a combination of CPUs — which tend to be less efficient for those sorts of tasks than GPUs — and a custom chip designed for accelerating AI algorithms. Meta pulled the plug on a large-scale rollout of the custom chip, which was planned for 2022, and instead placed orders for billions of dollars’ worth of Nvidia GPUs that required major redesigns of several of its data centers.

In an effort to turn things around, Meta made plans to start developing a more ambitious in-house chip, due out in 2025, capable of both training AI models and running them. And that was the main topic of today’s presentation.

Meta calls the new chip the Meta Training and Inference Accelerator, or MTIA for short, and describes it as a part of a “family” of chips for accelerating AI training and inferencing workloads. (“Inferencing” refers to running a trained model.) The MTIA is an ASIC, a kind of chip that combines different circuits on one board, allowing it to be programmed to carry out one or many tasks in parallel.

“To gain better levels of efficiency and performance across our important workloads, we needed a tailored solution that’s co-designed with the model, software stack and the system hardware,” Bjorlin continued. “This provides a better experience for our users across a variety of services.”

Custom AI chips are increasingly the name of the game among the Big Tech players. Google created a processor, the TPU (short for “tensor processing unit”), to train large generative AI systems like PaLM-2 and Imagen. Amazon offers proprietary chips to AWS customers both for training (Trainium) and inferencing (Inferentia). And Microsoft, reportedly, is working with AMD to develop an in-house AI chip called Athena.

Meta says that it created the first generation of the MTIA — MTIA v1 — in 2020, built on a 7-nanometer process. It can scale beyond its internal 128 MB of memory to up to 128 GB, and in a Meta-designed benchmark test — which, of course, has to be taken with a grain of salt — Meta claims that the MTIA handled “low-complexity” and “medium-complexity” AI models more efficiently than a GPU.

Work remains to be done in the memory and networking areas of the chip, Meta says, which present bottlenecks as the size of AI models grow, requiring workloads to be split up across several chips. (Not coincidentally, Meta recently acquired an Oslo-based team building AI networking tech at British chip unicorn Graphcore.) And for now, the MTIA’s focus is strictly on inference — not training — for “recommendation workloads” across Meta’s app family.

But Meta stressed that the MTIA, which it continues to refine, “greatly” increases the company’s efficiency in terms of performance per watt when running recommendation workloads — in turn allowing Meta to run “more enhanced” and “cutting-edge” (ostensibly) AI workloads.

A supercomputer for AI
Perhaps one day, Meta will relegate the bulk of its AI workloads to banks of MTIAs. But for now, the social network’s relying on the GPUs in its research-focused supercomputer, the Research SuperCluster (RSC).

First unveiled in January 2022, the RSC — assembled in partnership with Penguin Computing, Nvidia and Pure Storage — has completed its second-phase buildout. Meta says that it now contains a total of 2,000 Nvidia DGX A100 systems sporting 16,000 Nvidia A100 GPUs.

So why build an in-house supercomputer? Well, for one, there’s peer pressure. Several years ago, Microsoft made a big to-do about its AI supercomputer built in partnership with OpenAI, and more recently said that it would team up with Nvidia to build a new AI supercomputer in the Azure cloud. Elsewhere, Google’s been touting its own AI-focused supercomputer, which has 26,000 Nvidia H100 GPUs — putting it ahead of Meta’s.

But beyond keeping up with the Joneses, Meta says that the RSC confers the benefit of allowing its researchers to train models using real-world examples from Meta’s production systems. That’s unlike the company’s previous AI infrastructure, which leveraged only open source and publicly available datasets.

“The RSC AI supercomputer is used for pushing the boundaries of AI research in several domains, including generative AI,” a Meta spokesperson said. “It’s really about AI research productivity. We wanted to provide AI researchers with a state-of-the-art infrastructure for them to be able to develop models and empower them with a training platform to advance AI.”

At its peak, the RSC can reach nearly 5 exaflops of computing power, which the company claims makes it among the world’s fastest. (Lest that impress, it’s worth noting some experts view the exaflops performance metric with a pinch of salt and that the RSC is far outgunned by many of the world’s fastest supercomputers.)

Meta says that it used the RSC to train LLaMA, a tortured acronym for “Large Language Model Meta AI” — a large language model that the company shared as a “gated release” to researchers earlier in the year (and which subsequently leaked in various internet communities). The largest LLaMA model was trained on 2,048 A100 GPUs, Meta says, which took 21 days.

“Building our own supercomputing capabilities gives us control at every layer of the stack; from datacenter design to training frameworks,” the spokesperson added. “RSC will help Meta’s AI researchers build new and better AI models that can learn from trillions of examples; work across hundreds of different languages; seamlessly analyze text, images, and video together; develop new augmented reality tools; and much more.”

Video transcoder
In addition to MTIA, Meta is developing another chip to handle particular types of computing workloads, the company revealed at today’s event. Called the Meta Scalable Video Processor, or MSVP, the chip is Meta’s first in-house-developed ASIC solution designed for the processing needs of video on demand and live streaming.

Meta began ideating custom server-side video chips years ago, readers might recall, announcing an ASIC for video transcoding and inferencing work in 2019. This is the fruit of some of those efforts, as well as a renewed push for a competitive advantage in the area of live video specifically.

“On Facebook alone, people spend 50% of their time on the app watching video,” Meta technical lead managers Harikrishna Reddy and Yunqing Chen wrote in a co-authored blog post published this morning. “To serve the wide variety of devices all over the world (mobile devices, laptops, TVs, etc.), videos uploaded to Facebook or Instagram, for example, are transcoded into multiple bitstreams, with different encoding formats, resolutions and quality … MSVP is programmable and scalable, and can be configured to efficiently support both the high-quality transcoding needed for VOD as well as the low latency and faster processing times that live streaming requires.”

Meta says that its plan is to eventually offload the majority of its “stable and mature” video processing workloads to the MSVP and use software video encoding only for workloads that require specific customization and “significantly” higher quality. Work continues on improving video quality with MSVP using preprocessing methods like smart denoising and image enhancement, Meta says, as well as post-processing methods such as artifact removal and super-resolution.

“In the future, MSVP will allow us to support even more of Meta’s most important use cases and needs, including short-form videos — enabling efficient delivery of generative AI, AR/VR and other metaverse content,” Reddy and Chen said.

AI focus
If there’s a common thread in today’s hardware announcements, it’s that Meta’s attempting desperately to pick up the pace where it concerns AI, specifically generative AI.

As much had been telegraphed prior. In February, CEO Mark Zuckerberg — which has reportedly made upping Meta’s compute capacity for AI a top priority — announced a new top-level generative AI team to, in his words, “turbocharge” the company’s R&D. CTO Andrew Bosworth likewise said recently that generative AI was the area where he and Zuckerberg were spending the most time. And chief scientist Yann LeCun has said that Meta plans to deploy generative AI tools to create items in virtual reality.

“We’re exploring chat experiences in WhatsApp and Messenger, visual creation tools for posts in Facebook and Instagram and ads, and over time video and multi-modal experiences as well,” Zuckerberg said during Meta’s Q1 earnings call in April. “I expect that these tools will be valuable for everyone from regular people to creators to businesses. For example, I expect that a lot of interest in AI agents for business messaging and customer support will come once we nail that experience. Over time, this will extend to our work on the metaverse, too, where people will much more easily be able to create avatars, objects, worlds, and code to tie all of them together.”

In part, Meta’s feeling increasing pressure from investors concerned that the company’s not moving fast enough to capture the (potentially large) market for generative AI. It has no answer — yet — to chatbots like Bard, Bing Chat or ChatGPT. Nor has it made much progress on image generation, another key segment that’s seen explosive growth.

If the predictions are right, the total addressable market for generative AI software could be $150 billion. Goldman Sachs predicts that it’ll raise GDP by 7%.

Even a small slice of that could erase the billions Meta’s lost in investments in “metaverse” technologies like augmented reality headsets, meetings software and VR playgrounds like Horizon Worlds. Reality Labs, Meta’s division responsible for augmented reality tech, reported a net loss of $4 billion last quarter, and the company said during its Q1 call that it expects “operating losses to increase year over year in 2023.”

WSJ : Antitrust Gone Wild Against Amgen

Antitrust Gone Wild Against Amgen
No theory is too strange for Lina Khan’s FTC to block a merger.

Try though she has, Lina Khan hasn’t been able to stop every corporate merger. The frustration must be getting to her because the Federal Trade Commission Chair is resorting to ever more bizarre theories of antitrust to block business tie-ups.

On Tuesday the FTC sued to block Amgen’s acquisition of Horizon Therapeutics, which the agency boasts is the first pharmaceutical deal in “recent memory” that it has challenged. This is not an achievement to celebrate. Historically, the agency has raised competition concerns only when a drug maker has sought to buy a company with rival products. Then it has required discrete product divestments and approved the deals.

In this case, Amgen and Horizon don’t make drugs that directly compete. The FTC’s strained argument is that Amgen will use its negotiating clout with pharmaceutical benefit managers (PBMs) to protect Horizon’s successful drugs Tepezza and Krystexxa from competition. The drugs treat thyroid eye disease and chronic refractory gout, respectively.

The two drugs currently don’t have competition, and they may never get any. The Food and Drug Administration designated both “orphan drugs” because they treat rare diseases. This status confers regulatory benefits that are intended to provide an incentive to develop drugs for rare diseases for which there may be little financial return.

The FTC lawsuit concedes all this. “The preclinical and clinical trials can cost hundreds of millions of dollars to complete, all without a guarantee of success,” the lawsuit says. “The Department of Health and Human Services estimates that it can take $300-500 million and 14 years on average to develop and bring a drug to market.”

Yet the FTC claims that if a smaller company were to someday gain approval for a competing drug, Amgen might then offer PBMs larger rebates for its other drugs in return for giving Horizon’s treatments preferred placement on insurer formularies. It says this potential scenario might discourage potential competition.

The Khan FTC has been seeking to revive the ancient antitrust theory of “potential competition,” which was long ago shown to be flawed by antitrust legal scholars and has been discounted in the courts. In the Amgen case, the agency is taking an even greater logical leap in extending what qualifies as a potential antitrust violation.

The FTC doesn’t even attempt to show consumer harm, which is the modern standard for blocking mergers. If the scenario the FTC describes did occur, patients would probably benefit because PBMs use drug maker rebates to reduce insurance premiums.

To boost its weak case, the agency offers misdirection about how consolidation between insurers and PBMs in recent years could make its speculative case more likely. The FTC is investigating drugmaker rebates but hasn’t shown them to be illegal. So Ms. Khan is using the Amgen-Horizon lawsuit to discourage both rebates and acquisitions.

The lawsuit “sends a clear signal to the market: The FTC won’t hesitate to challenge mergers that enable pharmaceutical conglomerates to entrench their monopolies at the expense of consumers and fair competition,” the agency’s Bureau of Competition Director Holly Vedova said.

The FTC also says Amgen has built its drug portfolio by acquisition, which mimics Ms. Khan’s criticism of Big Tech. But this is how drug development in the U.S. works: Start-ups develop drugs because they are more nimble and have expertise in niche areas. Large drug makers then buy them because they are better at commercializing products and navigating the drug-approval obstacle course.

Investors fund start-ups on the expectation they will be acquired. Ms. Khan’s lawsuit will disrupt their calculations and hamper investment and innovation, especially in treatments for rare diseases. The Amgen-Horizon lawsuit may be Ms. Khan’s biggest and most destructive legal overreach so far, and apparently she thinks this is a compliment.

FT : Fund manager Amundi shifts assets from US to China

Fund manager Amundi shifts assets from US to China
Paris-based firm attracted by valuations and says investors have priced in ‘too much risk’

Europe’s largest asset manager Amundi is moving out of US assets in favour of China, attracted by the country’s brighter economic prospects, better valuations and a more benign outlook for inflation.

Vincent Mortier, chief investment officer of the Paris-based fund firm, which has €2.1tn in assets, thinks “too much risk” is priced into Chinese credit and high-quality companies, while markets in the US are “too optimistic” as a recession looms.

“In our allocations we have made a clear shift from west to east,” Mortier said in an interview with the Financial Times, predicting that the US economy will not grow next year while China, India and Indonesia will each grow by 5 to 6 per cent.

Mortier’s bullish stance on the world’s second-largest economy comes as political tensions between Beijing and Washington worsen and recent consumer and industrial activity has fallen far short of expectations.

In late January, a top US air force general predicted that the US and China would probably go to war in 2025, and the following month Chinese spy balloons were discovered over the US.

While Mortier sees geopolitics as “a risk”, he thinks the balance of power is such that the US would try to avoid being too harsh on China. Amundi has been gradually increasing its allocation to China and India over the past 12 months, and has accelerated the move this year.

“The US should not underestimate the capacity of China to retaliate, to escalate and then to negotiate,” Mortier said. “I think China can make serious arguments to put pressure on the US as well.” 

Mortier is particularly keen on opportunities in select Chinese corporate bonds, arguing that foreign investors have been blanket selling without differentiating between the quality of issuers.

“If you dig you can find really good companies that you can buy for 50 cents to the dollar,” he said. “If you are brave or if you have some time it’s very good.”

Amundi’s optimism on China comes despite a tough period for its equity markets. The CSI 300 Index has shed 5.7 per cent from its January peak.

Markets have been pushed lower by disappointing economic data. Although China’s economy grew by a better-than-expected 4.5 per cent year-on-year in the first quarter, industrial production rose 5.6 per cent last month from a year earlier, but well below forecasts of a 10.6 per cent rise. Retail sales figures also missed forecasts. But Mortier said he believed investors’ reaction may have been overblown.

“The data may seem disappointing but when you dig more into the details in fact they are not that bad,” he said. “I think the markets may have not caught the reality of what is happening because there have been changes in how the data is produced.”

US markets are pricing in a “goldilocks” scenario of low inflation, rate cuts, stronger earnings and a soft landing, but the likelihood of this happening is becoming increasingly remote as financial conditions tighten, he said.

The US Federal Reserve’s quarterly Senior Loan Officer Opinion Survey last week showed that 46 per cent of US banks plan to raise their lending standards owing to worries about loan losses and deposit flight.

Meanwhile, hourly wage growth was 4.4 per cent year on year in April, which is likely to put upward pressure on inflation.

Mortier thinks inflation in both the US and Europe will stabilise at around 3 to 4 per cent in the years ahead, but with some volatility that will make the Fed’s job particularly difficult owing to the time it takes for monetary policy to have an impact.

However, he is not worried about the possibility of a US debt default, which has sent the cost of insuring against US Treasury default soaring this year to its highest level since the financial crisis.

“We think Treasuries are safe,” he said. “We are continuing to manage this exposure like business as usual and even taking some opportunity” to buy some.

Nevertheless, he favours emerging markets over western markets as a whole and is upping allocation to both India and Indonesia where, like China, economic growth is becoming less dependent on exports.

“This is good news for these countries and less good news for the West,” he said.

FT : Grayscale aims for possible loophole with new bitcoin ETF filing

Grayscale aims for possible loophole with new bitcoin ETF filing
If approved, the US-listed vehicle will invest in ‘spot’ bitcoin exchange traded products trading in other countries

Grayscale, manager of the world’s largest cryptocurrency fund, has long been stymied by regulators in its bid to convert its flagship vehicle into an exchange traded fund.

However, it may have found a ruse to partially circumvent the restrictions imposed by the US Securities and Exchange commission.

Grayscale has filed to launch a US-listed ETF that would partly invest in so-called “spot” bitcoin exchange traded products already up and running in other countries.

These ETPs invest directly in “physical” bitcoin, or track the cryptocurrency’s price synthetically — structures the SEC has so far not permitted.

Instead, the SEC has only allowed bitcoin ETFs that trade in futures contracts listed on the Chicago Mercantile Exchange, which is a regulated venue.

This stance has prompted Grayscale to sue the SEC for its refusal to allow the $16.9bn Grayscale Bitcoin Trust (GBTC), a private spot bitcoin trust, to convert to an ETF. A ruling is due by the end of the third quarter.

In the interim, though, Grayscale has filed to launch a Global Bitcoin Composite ETF, which would invest 40 per cent of its assets in spot bitcoin ETPs listed on “major non-US exchanges”, such as in Canada or Europe.

While there is no guarantee that the application will be approved, analysts believe it has a chance given that the overseas ETPs Grayscale is proposing to invest in are regulated securities — precisely the type of asset that the SEC has insisted crypto ETFs hold.

“Grayscale’s filing seems intended to push the envelope with the SEC by using the SEC’s own words against it to launch crypto ETFs,” said Bryan Armour, director of passive strategies research, North America at Morningstar.

“This seems like Grayscale is directly challenging the SEC’s past decision to allow bitcoin futures ETFs on one hand and on the other hand reject spot bitcoin ETFs,” he added.

Todd Rosenbluth, head of research at VettaFi, a consultancy, described the filing “as a creative way of Grayscale trying to offer US-listed product with some spot bitcoin exposure”.

“I think there is greater likelihood of Grayscale getting approval for this ETF than it getting the SEC to voluntarily approve a US-listed spot bitcoin only ETF, which is what Grayscale wants,” he added.

Rosenbluth believed the filing was an “acceptance of the reality that the SEC is not going to give [Grayscale] the green light”, to convert GBTC.

One point in Grayscale’s favour in its latest filing is that vast numbers of US-listed ETFs already hold regulated, non-US securities, mostly in the form of equities and bonds.

Indeed, at least one US ETF already holds foreign-listed spot bitcoin ETFs, appearing to set something of a precedent for the structure.

The $447mn Amplify Transformational Data Sharing ETF (BLOK) has positions in four Toronto-listed spot bitcoin ETFs, although they currently account for just 4.3 per cent of its portfolio, with the balance in stocks of companies involved in blockchain technology.

Grayscale is proposing that its 40 per cent exposure to spot bitcoin ETFs would be weighted equally between five underlying funds.

Rosenbluth also noted the “added irony” of Cathie Wood’s Ark Next Generation Internet ETF (ARKW) having a 6.8 per cent exposure to Grayscale’s GBTC trust, even though the SEC deems GBTC’s structure to be unsuitable for an ETF. The SEC declined to comment.

Even if Grayscale was to get approval for the Global Bitcoin Composite ETF, it is far from certain there would be a huge appetite for the fund, however.

It is proposing that the remaining 60 per cent of the portfolio would be invested in the equity of bitcoin mining companies. Armour was unconvinced there was much demand for such a structure.

“I don’t think investors benefit from access to a 60/40 portfolio of bitcoin miners and a global spot bitcoin fund of funds sleeve,” he said.

“The portfolio is confusing and will do a worse job tracking the spot bitcoin price than a bitcoin futures ETF, so I’m not sure this type of product adds any value aside from parsing out the SEC’s strategy for approving bitcoin-related ETFs.”

Rosenbluth agreed demand might be lacking, saying “combining bitcoin with other investments is not novel but has had limited interest”.

As evidence he cited the Global X Blockchain & Bitcoin Strategy ETF (BITS), which invests in the Global X Blockchain ETF (BKCH) and bitcoin futures, but which holds just $11mn. 

Rosenbluth noted that despite a strong bounce for bitcoin and crypto-related companies after last year’s sharp losses, with the currency itself up 61 per cent so far this year, “demand for ETFs exposed to bitcoin has been limited”.

“Investors seem sceptical the rally can persist and remember the challenges of 2022,” he added.

For instance, although the $1bn ProShares Bitcoin Strategy ETF (BITO), the largest bitcoin futures ETF, was up 57 per cent year to date as of May 12, it had only seen net inflows of $95mn, according to VettaFI’s data. This was not far ahead of the $75mn attracted by the $120mn ProShares Short Bitcoin ETF Strategy (BITI), which takes the opposite bet and is down 44 per cent year to date.

BKCH has returned 69 per cent so far this year but has seen net outflows of $2mn. Likewise the Invesco Alerian Galaxy Crypto Economy ETF (SATO) is up 68 per cent “and has seen no investor interest either”, Rosenbluth added.

Michael Sonnenshein, chief executive of Grayscale, was more bullish, however, arguing that “there is no question, coming out of the crypto winter, that investors’ appetite for crypto remains healthy. It has certainly not gone unnoticed by the crypto community that crypto has achieved some of the highest returns in 2023 to date.”

>>> US After Hours Summary: FTCH +16.8%, PLXS +0.7% higher on earnings/guidance;

After Hours Summary: FTCH +16.8%, PLXS +0.7% higher on earnings/guidance; DXC -5.7%, FLO -4.9%, AMAT -1.5% lower on earnings

After Hours Gainers:

Companies trading higher in after hours in reaction to earnings/guidance: FTCH +16.8%, PLXS +0.7%, GLOB +0.2%

Companies trading higher in after hours in reaction to news: CPE +1.6% (names new COO), KAI +0.6% (authorizes new $50 mln share repurchase program), GD +0.5% (awarded $490 mln U.S. Army contract), ADI +0.2% (CFO to step down, reaffirms its Q2 guidance), LMT +0.2% (awarded $750 mln and $444 mln US Air Force contracts), SAIC +0.1% (CEO to retire, names new CEO), NSC +0.1% (reaches deal with union on paid sick leave)

After Hours Losers:

Companies trading lower in after hours in reaction to earnings/guidance: DXC -5.7% (also names new CFO; also authorizes new $1 bln share repurchase program), FLO -4.9%, CVCO -3.9%, AMAT -1.5%, ROST -0.1%

Companies trading lower in after hours in reaction to news: UMH -2% (files mixed securities shelf offering), AMAM -1.9% (names new chairman), RDN -1.7% (names new CFO), ZG -0.3% (names new CFO), FTI -0.1% (awarded significant contract by Equinor), MU -0.1% (to bring extreme ultraviolet (EUV) tech to Japan)

FT : Carl Icahn admits he got bearish bet that cost $9bn wrong

Carl Icahn admits he got bearish bet that cost $9bn wrong
‘If I kept the parameters I always believed in . . .  I would have been fine,’ concedes activist

Carl Icahn has admitted he was wrong to make a huge bet that the market would crash after the ill-fated trade cost his firm nearly $9bn over roughly six years.

According to a Financial Times analysis, the prominent activist investor lost about $1.8bn in 2017 on hedging positions that would have paid out if asset prices had tumbled before losing a further $7bn between 2018 and the first quarter of this year.

“I’ve always told people there is nobody who can really pick the market on a short-term or an intermediate-term basis,” Icahn told the FT in an interview to discuss the analysis. “Maybe I made the mistake of not adhering to my own advice in recent years.”

Icahn Enterprises started aggressively betting on a market collapse in the aftermath of the 2008 financial crisis and became increasingly bolder in subsequent years, deploying a complex strategy that involved shorting broad market indices, individual companies, commercial mortgages and debt securities.

At times, Icahn’s notional exposure, the underlying value of the securities he was betting against, exceeded $15bn, regulatory filings show. “You never get the perfect hedge, but if I kept the parameters I always believed in . . . I would have been fine,” he said. “But I didn’t.”

Icahn Enterprises, the listed vehicle majority owned by the activist that allows retail investors to join in his wagers, reported a total of $4.3bn in short losses in 2020 and 2021 as markets quickly rebounded from the pandemic slump following the Federal Reserve’s huge stimulus.

“I obviously believed the market was in for great trouble,” said Icahn. “[But] the Fed injected trillions of dollars into the market to fight Covid and the old saying is true: ‘don’t fight the Fed’.”

The trades have left Icahn in a vulnerable position and threaten to undermine his status as one of the most feared activist investors on Wall Street.

Earlier this month, short seller Hindenburg Research released a report saying it believed the market value of Icahn Enterprises was inflated and its dividend was unsustainable. Shares of the company have fallen by more than 30 per cent since the report was published.

As Icahn’s short bets drained billions of dollars from his investment firm, he ploughed nearly $4bn of his own money into his publicly listed vehicle, filings show. That injection helped keep the firm’s internally calculated investment portfolio value relatively stable.

Icahn exposed himself to another risk by taking out a margin loan that was first disclosed in early 2022. Hindenburg’s report drew attention to the margin loan from Morgan Stanley, against which Icahn pledged 60 per cent of his stake in Icahn Enterprises as collateral.

Hindenburg argued this could lead to his business unravelling if the plunging stock price triggers a margin call that would force Icahn to liquidate some of his stake.

In a statement earlier this month that addressed Hindenburg’s allegations, Icahn Enterprises said Icahn was in “full compliance” with regards to all personal loans and announced a $500mn stock buyback authorisation in a bid to bolster its share price. With regards to its market valuation, the company said that “over time, [our] performance will speak for itself”.

Icahn told the FT that he had used the margin loan to make additional investments and had billions of dollars of cash outside of his public vehicle. “Over the years I have made a great deal of money with money,” he said. “I like to have a war chest and doing that gave me more of a war chest,” he added, referring to the margin loan.

Icahn Enterprises has warned that “a prolonged decline” in its stock price “could increase the likelihood of a foreclosure or forced sale” of Icahn’s stake if he was “subject to a ‘margin call’”.

Earlier this month, Icahn Enterprises revealed federal prosecutors in New York had contacted the company seeking information on its business, including corporate governance, valuations and due diligence.

Icahn’s bearish bets are the main reason his investment portfolio has lost money in every year since 2014. Over the roughly six year period that he lost $9bn on the short bets, the portfolio made about $6bn from his activist wagers, leaving the vehicle with an overall investment loss of nearly $3bn.

Separately, Icahn Enterprises generated $3.5bn of gains during the period by selling companies it controlled — including casinos and a railcar leasing business — that were held outside the investment portfolio.

The net asset value of Icahn Enterprises fell from $7.9bn in 2017 to $5.6bn this month. That poses a potential problem for Icahn, who has historically taken the large $8 a share annual dividend in stock rather than cash. This has caused the number of outstanding shares to more than double over the roughly six-year period, pushing its net asset value per share down from $33 to roughly $16.

Retail investors who took their dividends in cash would have received more than $40 a share during the same period.

As pressure on his firm mounts, Icahn has been forced to rein in his short bets just as some investors fear that a regional banking crisis and the debt ceiling stand-off could result in a sharp market sell-off.

“I still to some extent believe that this economy is not good and there are going to be problems ahead,” Icahn said. “We are still hedged, but not to the extent we were.”

>>> US Gapping down

Gapping down
In reaction to earnings/guidance:

  • BOOT -13.6%, LSPD -13.2%, CSCO -3.7%, MNRO -3.4%, SQM -3%, GRAB -2.8%, NGG -2.4%, STNE -2.3%, VSAT -2%, BABA -1.5% (also approves spin-off)

Other news:

  • NSTG -8.3% (TXG wins injunction in patent litigation with NSTG)
  • APLD -5.7% (files $175 mln mixed shelf securities offering)
  • GFL -3.5% (announces upsizing and pricing of secondary offering by selling shareholders)
  • EVGO -3% (prices offering of 29411765 shares of common stock at $4.25 per share)
  • OUST -2.6% (confirms the ITC has voted to institute an investigation into the unfair trade practices of Hesai Group (HSAI) based in Shanghai China and related entities)
  • MNMD -2.2% (announces enrollment milestone in Phase 2b trial of MM-120)
  • KALU -2.1% (Warrick mill union members ratify labor agreement)
  • FRPT -2% (advances operational improvement plan through Board changes; Current Director Walt George will be named Chair to replace retiring Charles Norris)

Analyst comments:

  • WTRG -1.5% (downgraded to Neutral from Buy at Northcoast)
  • MMP -1.2% (downgraded to Hold from Buy at Stifel)

>>> US Gapping up

Gapping up
In reaction to earnings/guidance:

  • DLO +15.7%, GOOS +13.5%, TTWO +12.8%, BBWI +9.5%, BEKE +6.6%, DOLE +4.6%, CSIQ +4.5%, ZTO +3.9%, EXP +2.4%, BOWL +2%, SNPS +1.9%, WMT +1.4%

Other news:

  • EYPT +10.5% (sold YUTIQ (fluocinolone acetonide intravitreal implant) 0.18mg to Alimera Sciences (ALIM))
  • SONY +4% (assesses partial spin-off)
  • CABA +3.8% (prices offering of 7.25 mln shares of common stock at $12.00 per share)
  • ACET +3% (Preclinical Data on ADI-270 at the American Society of Gene and Cell Therapy (ASGCT) 26th Annual Meeting)
  • TXG +2.8% (TXG wins injunction in patent litigation with NSTG)
  • SNOW +2.2% (in talks to buy search startup Neeva according to The Information)
  • UBA +2% (Regency Centers (REG) to acquire Urstadt Biddle Properties in all-stock transaction valued at approximately $1.4 bln)
  • FGEN +2% (topline data from Company's Phase 3 clinical study of roxadustat for treatment of anemia in patients receiving concurrent chemotherapy treatment for non-myeloid malignancies in China)
  • PLX +1.7% (is eligible to receive a $20 million milestone payment from its commercial partner Chiesi Global Rare Diseases)

Analyst comments:

  • ALB +2% (upgraded to Buy from Neutral at UBS)
  • DT +1.8% (upgraded to Buy from Neutral at BTIG Research)

ok

>>> US Early premarket gappers

Early premarket gappers

  • Gapping up:
    • DLO +16.4%, TTWO +11.8%, BBWI +6.5%, CARA +4.6%, MNMD +4.4%, SONY +4.3%, EXP +4.2%, DOLE +4%, ZTO +3.9%, CSIQ +3.3%, ACET +2.1%, BOWL +2%, SNOW +1.6%, SNPS +1.6%, KD +1.1%, SAP +0.9%, BEKE +0.9%, HOOD +0.8%, CABA +0.7%
  • Gapping down:
    • BOOT -14%, NSTG -7.1%, APLD -5%, CSCO -4.2%, OUST -3.9%, SQM -3.6%, GFL -3.2%, STNE -3.1%, VSAT -2.5%, NGG -2.2%, KALU -2.1%, FRPT -2%, EVGO -1.9%, LITE -1.4%, GTES -1.3%, TXG -1.2%, CB -1%, JNPR -0.8%, AVGO -0.7%, BWMN -0.7%