After an intense and eventful day at Goldman Sachs’ highly anticipated technology conference earlier this week, I found myself seated with the firm’s well-regarded internet analyst, Eric Sheridan, in order to reflect on the overwhelming flood of insights, perspectives, and strategic revelations that had emerged during the various sessions. His most pressing and distilled conclusion could not have been stated more clearly: the technology industry shows no indication whatsoever of reducing its pace of ambition or innovation. In his words, “No one’s slowing down,” and judging from the dominant themes of the event, the sentiment carried little exaggeration.
Although the sector is already committing enormous sums of money toward the rapid construction of artificial intelligence infrastructure, Sheridan explained that nearly every senior executive in attendance emphasized a recurring challenge: the appetite for AI capability is escalating faster than even the most well-funded companies can meet. This mismatch between supply and demand has become an urgent narrative. Nowhere was this more apparent than in comments from leaders at CoreWeave, a company specializing in the design and operation of highly advanced AI data centers. They described the demand with a single, almost visceral word—“unrelenting.” Their remarks conveyed that, in the last month to month and a half, there has been yet another sharp upward shift in how intensely organizations are seeking AI access and computational power.
Sheridan offered a historical comparison that illuminated why this moment feels distinct. During the dot-com boom of the late 1990s, vast amounts of internet infrastructure were built seemingly on the speculative promise of “eyeballs”—a term that represented little more than website traffic or audience attention. By contrast, the current AI wave is grounded in something far sturdier than optimistic projections. This time around, there are tangible and substantial flows of revenue: individuals and businesses are actively paying for AI tools, platforms, and products. That essential difference makes this expansion feel far more sustainable and consequential.
The marquee presentation of the conference was delivered by Sarah Friar, the Chief Financial Officer of OpenAI. The interest she drew was extraordinary; the primary hall was completely filled, and the overflow room, intended merely as backup, was also packed to capacity. Analysts, investors, and technologists crowded together, many resorting to sitting cross-legged on the floor. The sight of suited financiers, polished shoes tucked beneath them, symbolized just how unusual and magnetic the moment has become.
Friar disclosed impressive financial figures: OpenAI is on track to generate approximately $13 billion in revenue this year. Yet, she was equally candid about the formidable constraints the organization continues to face. Despite monumental progress, the company remains “massively compute constrained.” As a result, executives are forced to make difficult strategic decisions, which include intentionally delaying the launch of new products, purposefully slowing down certain services in order to conserve computational resources, and prioritizing among research projects, some of which must be postponed until greater capacity is secured.
Sheridan emphasized that such limitations are reshaping alliances in unexpected ways, creating what he described as “strange bedfellows.” At the same event, Meta’s CFO Susan Li revealed that the social media giant is actively cooperating with Google—an institution often seen as its bitter competitor. Likewise, Friar admitted that OpenAI is relying on Google’s cloud infrastructure to bolster its computing capacity. These partnerships arise even as the very same companies battle intensely over markets as strategically vital as AI-assisted search.
However, amid the optimism, there was a shadow that lingered over the gathering. The worrisome theme that unsettled some participants was the broader disruption that AI might unleash upon the software industry. This anxiety is already exerting downward pressure on the valuations of software-as-a-service providers. When questioned directly about the subject, Friar did not choose to soften her response. She explained that in an environment where autonomous software development becomes standard, the logic behind procuring generic software solutions weakens. When organizations are empowered to create precisely tailored programs internally with relative ease, the incentive to purchase third-party software diminishes significantly. Offering a rhetorical question, she asked: “Why wouldn’t I build the exact kind of software that my company needs?” Her conclusion suggested nothing less than a fundamental reshaping of the software development landscape.
That observation reverberated throughout the audience. As listeners absorbed the potential magnitude of the changes Friar described—the possibility of AI consuming vast swaths of the global software ecosystem—a palpable unease hung in the air. At one point, a voice from nearby muttered darkly, “Short everything,” a quip that was half jest yet tinged with genuine concern. The remark drew nervous laughter, and as the audience slowly rose and shuffled out, the mood was one of contemplative uncertainty. The line of attendees moving toward the exits was uncharacteristically subdued, each person considering what the acceleration of AI might mean for industries, investments, and the very structure of technological progress.
Sourse: https://www.businessinsider.com/ai-message-from-silicon-valley-no-one-slowing-down-2025-9
