spot_imgspot_img

Top 5 This Week

spot_img

Related Posts

Anthropic debuts to skeptics and dogged believers: Mike Dolan



Six weeks of jitters among AI investors are now colliding with the market debut of one of the sector’s biggest champions. For all the fanfare around its share sale, Anthropic’s arrival is calming no one down.

The Claude developer’s bumper initial public offering (IPO) is set to price on Wednesday before trading begins the following day. In the rush to go public before rival ChatGPT owner OpenAI, it has been caught in a volatility storm that’s seen some $2 trillion wiped off AI chip giant and ubiquitous AI financier Nvidia’s market value in just six weeks.

What was touted only a couple of months ago as the world’s biggest IPO, with a $2 trillion valuation, is now guided at least 30% lower than that. It also looks well short of SpaceX’s valuation as recently as June. SpaceX, for what it’s worth, is trading just 1% above its issue price five months ago.

Even though this month’s ouster of Nvidia boss Jensen Huang on revelations of multiple extramarital affairs and allegations of spying for China can explain some of its hit, the angst was underway well before the Huang saga hit.

With worries about political pushback against AI data centers following the US midterm elections rattling sentiment, Wall Street’s Nasdaq 100 is down 12% in that time, with the leading US chipmaker index down more than 25%.

The Cboe Volatility Index “fear index” of implied US equity market volatility is creeping higher toward levels that shout: “Avoid big IPOs.”

Sudden concerns about AI are clearly not about Anthropic’s numbers per se. The firm’s prospectus and the many bankers desperate to sell the new stock are doing their job of sketching out its explosive growth and its humanity-threatening potential. It’s not even about a rethink of the blistering spend on the wider AI buildout either.

Despite the bruising shakeout of the past month, there seems to be little or no dent in estimates for capital expenditure flowing into the sector next year and beyond.

Morgan Stanley’s post-election tallies and 2027 lookaheads, for example, still estimate $1.6 trillion and $1.8 trillion of so-called hyperscaler capex for 2027 and 2028 respectively.

While the bank flags some shift in data center rollout from 2028 to 2029 as a result of the midterm wins in local and congressional voting, its top-line numbers for the next two years are unchanged from prior estimates.

It’s not hard to see why some think Anthropic can round up its believers and surf through the storm to a stellar future.

Or might it get chewed up in public before its rogue bots get a chance to kill us all?

Question time

Most of the answers lie somewhere in the endless back-and-forth about circular financing of the AI buildout, doubts about diffusion of sophisticated AI models throughout the wider economy, and the conundrums about compute costs and token pricing.

Credit rating firm Egan-Jones on Friday sketched out the pros and cons of Anthropic’s position.

The pros will be familiar to anyone listening to the marketing and cheerleading around the roadshow, despite clear market disappointment at the pace of its breakneck revenue updates over recent weeks.

The cons make for sober reading, pointing to several major flags and unknowns.

For a start, the ratings firm talks of possible capital overstretch, with close to $900 billion of purchase and lease commitments earmarked through 2045, about half of them due by 2030.

That’s currently almost four times its combined cash position, trailing revenue and share sale proceeds, and packs in a lot of hope about what’s coming down the pike.

That may be OK if the company has firmly baked-in revenue streams, but many of its top customers have no long-term contracts and can reduce their spending, generated overwhelmingly by consumption-based fees, pretty much any time they want.

Egan-Jones went on to spotlight the firm’s external reliance on companies that are already, or will ultimately be, its rivals.

“About 44% of revenue came through cloud marketplaces, where Amazon, Google and Microsoft stand between Anthropic and the buyer and sell competing models,” it said.

Much of the rest of its “cons” sketches significant vulnerabilities to competition within an AI ecosystem that it’s relied on for financial backing. It also highlights the risk of easy customer switching as product costs rise and open-weight or cheaper open-source models become more compelling.

And that presupposes there is significant economy-wide take-up of the more expensive models at all.

The glare of quarterly reporting and daily share trading may pressure Anthropic into more transparency about token use and profitability, potentially adding clarity to an opaque world. But as the cost of AI to end users becomes clearer and more closely tied to use of frontier models, it may shape what firms are willing to use and pay for.

“There are still massive questions left unanswered about Anthropic,” the report concluded.

Much like Nvidia, which is due to report quarterly results after Tuesday’s closing bell, the questions may just be individual corporate quirks and conundrums.

But the more interesting issue is whether the many doubts reflect an AI story that’s reliant on a punt in the dark rather than hard calculations about the future.

Do all tech transformations have to be like this?

Axa Group chief economist Gilles Moec this week dug deep into other tech booms, such as the electricity rollout in the 1920s or the telecoms scramble in the 1990s, and found familiar patterns of circular or vendor financing and infrastructure financing.

But he warned that you don’t need to have a black-and-white view of the boom to appreciate that markets can miscalculate the speed and trajectory of the returns built in, with regulatory changes a huge potential factor in market repricing or corporate success.

“Finding reassurance in circular demand…can be misleading,” Moec wrote. “The right metric is demand from the rest of the economy, for which precise data is, at this stage, scarce, and the outlook difficult to quantify.”

With an IPO of Anthropic’s sheer scale and importance to the existing technology at stake, faith alone feels like a feeble foundation.

(The opinions expressed here are those of Mike Dolan, a columnist for Reuters.)



Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles