Saturday, July 1, 2023

"Nvidia’s Next Big Opportunity Is AI-as-a-Service. Here’s Why." (NVDA)

A quick hit from Barron's June 30, 1:01 pm ET:

Raymond James said Nvidia ’s new artificial intelligence cloud offering for corporations could become a large business over time.

Analyst Srini Pajjuri reaffirmed on Thursday his Strong Buy rating for Nvidia stock (ticker: NVDA) and his $450 price target.

Earlier this year, Nvidia unveiled its DGX cloud service, which allows companies to rent AI computing capacity powered by the chip maker’s semiconductors.

Nvidia DGX Cloud’s AI-as-a-Service is a significant long term opportunity, he wrote. The chip maker has AI “models that enterprises can leverage to build customized large language models with proprietary enterprise data.”....


So, we are faced with the decision whether-or-not to play a dangerous little game, riding the bubble knowing full well it is a bubble, or retiring to the sidelines.
For now one of our favorite economists with one of our favorite stories.

Here's the version hosted at MIT:
This paper presents a case study of a well-informed investor in the South Sea bubble. We argue that Hoare’s Bank, a fledgling West End London bank, knew that a bubble was in progress and nonetheless invested in the stock: it was profitable to “ride the bubble.” Using a unique dataset on daily trades, we show that this sophisticated investor was not constrained by such institutional factors as restrictions on short sales or agency problems...
The two most important parts of the paper "II. Hoare’s Trading Performance" and "III. Causes of Success" are definitely worth a couple minutes.

Also posted in 2012 as "Prop Trading the South Sea Bubble: Hoare's Bank 1720":

What allows asset price bubbles to inflate? The recent rise and fall of technology stocks have led many to argue that wide swings in asset prices are largely driven by herd behavior among investors. Robert J. Shiller (2000) emphasized that “irrational exuberance” raised stock prices above their fundamental values in the 1990s. Others, however, have pointed to structural features of the stock market, such as lock-up provisions for IPOs, analysts’ advice, strategic interactions between investors, and the uncertainties surrounding Internet technology, as causes of the recent bubble. We use a historical example to ask which of these explanations is most apt, with the potential to shed light on other important episodes of market overvaluation.

We examine one of the most famous and dramatic episodes in the history of speculation, the South Sea bubble. Data on the daily trading behavior of a goldsmith bank—Hoare’s—allow us to examine competing explanations for how bubbles can inflate.

While many investors, including Isaac Newton, lost substantially in 1720, Hoare’s made a profit of over £28,000, a great deal of money at a time when £200 was a comfortable annual income for a middle-class family (John Carswell, 1993). The behavior of a single knowledgeable investor can tell us much about the nature of bubbles and investors during periods of substantial mispricing. The bank did not profit simply by chance. It “rode the bubble” for an extended period while giving numerous indications that it believed the stock to be overvalued. Short-selling constraints and the difficulties of arbitrage that have been emphasized in recent work on the dot-com mania cannot explain the South Sea bubble. A zero-investment constraint, if it existed, did not bind market participants like Hoare’s. Perverse incentive effects arising from delegated investment management highlighted in recent work on mutual funds and hedge funds were not at work. We infer that the need for coordination in attacking the South Sea bubble was the key to allowing it to inflate to such an extreme extent, in line with recent theoretical work by Dilip Abreu and Markus Brunnermeier (2003).

There is a rich body of earlier research on the emergence of bubbles. The efficient markets hypothesis rules out substantial mispricing (Eugene F. Fama, 1965). The same conclusion emerges from no-trade theorems under asymmetric information, as well as from backward induction in finite horizon models (Manuel Santos and Michael Woodford, 1997; Jean Tirole, 1982). Famous historical episodes like the South Sea bubble, the tulip mania, and the Mississippi speculation are given as examples of markets functioning reasonably well under uncertainty (Peter M. Garber, 2000). Recent theoretical and empirical work, however, suggests that bubbles can inflate even if there are large numbers of highly capitalized, rational investors....