Consider the audacity of the claim for a moment. A finance professional posts to X on a Tuesday, tells the world they spent one afternoon building a functional replica of the Bloomberg Terminal using an AI tool, and attracts 7.5 million views before most of their colleagues have finished their morning coffee. In the attention economy, that is a detonation. In the real economy, it deserves a great deal more scrutiny.
The tool at the centre of the story is Perplexity Computer, launched last week by the AI search company Perplexity AI. Perplexity, valued at $20 billion, describes Computer as a multi-model agent orchestration platform that coordinates 19 different AI models to complete complex, long-running workflows entirely in the background. The company says it is a computer user agent that can execute complex workflows independently using 19 different AI models, even creating subagents to handle specific problems. The tool is available only on the company's highest subscription tier, the $200-per-month Perplexity Max.
Almost every techie put it to the test, including X user @hamptonism (Hampton), who claimed to have needed but one afternoon to build a clone of Bloomberg's Terminal, going as far as saying "Perplexity just became the first AI company to truly go head-to-head with the Bloomberg Terminal." The post went viral. The hyperbole flowed freely. More than a few commentators declared Bloomberg finished.
Strip away the talking points and what remains is a question worth asking carefully: what does the Bloomberg Terminal actually do, and how much of it did this demonstration genuinely replicate?
What Bloomberg Actually Sells
For decades, the Bloomberg Terminal has been the undisputed operating system of global finance. Walk onto any trading floor in New York, London, or Hong Kong and you will see it glowing in black and amber. It is more than software; it is infrastructure. The Terminal does not just provide data; it provides identity. The Terminal is Bloomberg's bread-and-butter, representing about 85 per cent of the firm's approximate $12 billion yearly revenue.
Bloomberg's own figures show its service covers "more than 200 billion pieces of financial data daily, across 6.5 million entities", with feeds licensed directly from exchanges, regulators, and thousands of financial institutions. The Terminal is also a highly complex, highly customisable product, even coming with its own keyboard. The institutional messaging network alone, connecting traders and analysts across every major market, took decades to build and cannot be conjured from a chatbot prompt.
Once a company pays tens of thousands of dollars per year per employee for the Bloomberg Terminal, trains everyone on it, and builds workflows around it, they are locked in. Even if something slightly better comes along, the pain of switching is too high. This is the moat that has kept Bloomberg dominant through every previous wave of technological disruption, from the internet to cloud computing.
Where the Demo Falls Short
Many users immediately pointed out that the data feed for Hampton's clone comes from Perplexity Finance, which itself is an AI bot that aggregates information from various sources. This means the clone's "real-time" information is not quite so, even assuming that Perplexity Finance gets actual zero-delay, true real-time data to begin with, a doubtful fact on its own.
That is a feat Hampton's clone almost certainly cannot claim, whether for speed or for breadth of information. For retail investors doing casual research on a publicly listed company, this gap may be irrelevant. For a debt trader executing a position in the milliseconds after a central bank announcement, it is the entire ballgame. The demonstration was impressive as a proof of concept. As a like-for-like comparison, it was something else entirely.
Perplexity had invited the press to a background briefing with executives last week to discuss the product and lay out the agenda for the year. The event was intended to include a demonstration of the tool, but the company cancelled the demo because of flaws found in the product hours before the event. That detail has received considerably less attention than the viral post.
The fundamental question is not whether Perplexity Computer is impressive. It clearly is. Perplexity Computer is a system that creates and executes entire workflows, capable of running for hours or even months. A user can describe a desired outcome and Computer will autonomously break that project into components, assign each to the right model, and work on it in the background, checking in only when it genuinely needs input. These are genuinely useful capabilities. But useful is not the same as equivalent.
The Counter-Argument Deserves Serious Consideration
Those quick to dismiss the demo entirely are making their own analytical error. Independent traders, smaller funds, fintech startups, and global operators often need similar insights at a fraction of the budget. If a roughly $200-per-month AI subscription can replicate even a majority of key workflows, the pressure starts at the edges. Disruption rarely begins by replacing the incumbent's largest clients. It begins by serving those priced out.
This matters for Australian financial services in particular. The Australian Securities and Investments Commission has long noted the cost barriers smaller financial intermediaries face in accessing institutional-grade data and research tools. If AI genuinely democratises access to that kind of analytical capability, the benefits to market competition and consumer outcomes could be real. A boutique fund manager in Brisbane or Adelaide competing against the major banks deserves access to tools commensurate with their talent, not just their budget.
Bloomberg has proprietary datasets, specialised tools, compliance features, and decades of institutional trust. None of those things evaporate because a finance techie had a productive Tuesday afternoon on X. But the direction of travel is becoming harder to ignore, and institutions that assume their competitive moats are permanent should probably stop assuming.
The big risk for Perplexity, however, is if the models themselves become commodities, rendering useless a service that allows users to switch between the AIs for tasks. Perplexity's basic product repackaging has also created a significant problem for the company, with several copyright lawsuits pending against it. These are not minor concerns. The company's valuation depends on its orchestration layer being more than a thin wrapper around other people's models.
What This Moment Actually Reveals
History will judge this moment not by the accuracy of a single viral post, but by whether the underlying dynamic it pointed to proved durable. The Bloomberg Terminal is not going to be replaced by a $200 subscription this quarter, or perhaps this decade. Its data relationships, its compliance architecture, and its network of users are genuinely irreplaceable in the short term.
What is changing is the cost of building something that looks, to a growing number of users, close enough. It can be argued that with enough effort, folks can make a simplistic, skin-deep version of a Bloomberg Terminal that is enough for their needs. For the industry's most sophisticated participants, skin-deep is nowhere near good enough. For the vast majority of people who have never been able to afford the real thing, it may be transformative.
Reasonable people can disagree about how fast this transition will accelerate. What they should not do is let the noise of a viral post substitute for that harder analytical work. The Reserve Bank of Australia and financial regulators globally will eventually need to grapple with what AI-generated financial data means for market integrity. The Australian Parliament is only beginning to develop the legislative frameworks that will govern agentic AI systems operating in sensitive domains.
The claim that AI killed the Bloomberg Terminal in an afternoon was an overstatement. The claim that nothing significant happened last week would be an equally large one. The truth, as it usually does, sits somewhere in the complicated middle, which is precisely where serious analysis needs to begin.