Significant challenges are being faced by companies like OpenAI as they encounter scaling plateaus in developing new large language models. Host Dave Sobel notes that AI researchers are advocating for innovative training techniques that mimic human reasoning, suggesting a shift in the competitive landscape. OpenAI's latest model, O1, employs a technique called test-time compute, which allows for real-time evaluation of multiple possibilities, potentially enhancing performance and altering the demand for AI hardware, particularly NVIDIA's chips.
The episode also delves into the mixed sentiments surrounding AI-capable PCs. Despite a notable 49% growth in sales, these devices only represent 20% of the market, with many channel partners expressing skepticism about their value. A significant portion of IT decision-makers remains hesitant to trust AI for autonomous decision-making, reflecting broader concerns about security and ethical implications. Sobel emphasizes the need for clearer value propositions and better education to promote AI-capable PCs, especially as the industry transitions from the hype phase to a more normalized adoption of AI technologies.
Sobel highlights the increasing focus on cybersecurity as CrowdStrike launches its AI Red Team services to protect AI systems from rising cyber threats. With the rapid adoption of AI, risks such as model tampering and data poisoning are becoming more prevalent. The episode also touches on Microsoft's recent issues with unexpected upgrades to Windows Server, which have left some users in a difficult position, and the company's efforts to enhance its update procedures.
Finally, the episode addresses the energy challenges faced by tech companies, particularly Amazon, which had its request for increased power supply denied by regulators. As demand for data center power is projected to rise significantly, Sobel discusses the potential for partnerships with nuclear and renewable energy providers to ensure stability and sustainability in energy supply. The episode concludes by emphasizing the importance of energy efficiency in data center operations as companies navigate regulatory pressures and seek innovative solutions to meet their energy needs.
Four things to know today
00:00 OpenAI and AI Industry Pivot as Scaling Plateaus: Shift Towards Human-like Reasoning Could Redefine Market Competition
05:21 AI's Channel Opportunity Grows, But Skepticism Toward AI-Capable PCs Highlights Industry Trust Gap
09:01 CrowdStrike, Microsoft, and Box Drive AI and Security Enhancements Amid Rising Enterprise Demand
13:01 Amazon’s Power Request Denial Signals Rising Challenges in Data Center Energy Demands
Supported by: https://www.coreview.com/msp
https://getthread.com/mspradio/
💼 All Our Sponsors
Support the vendors who support the show:
👉 https://businessof.tech/sponsors/
🚀 Join Business of Tech Plus
Get exclusive access to investigative reports, vendor analysis, leadership briefings, and more.
👉 https://businessof.tech/plus
🎧 Subscribe to the Business of Tech
Want the show on your favorite podcast app or prefer the written versions of each story?
📲 https://www.businessof.tech/subscribe
📰 Story Links & Sources
Looking for the links from today’s stories?
Every episode script — with full source links — is posted at:
🎙 Want to Be a Guest?
Pitch your story or appear on Business of Tech: Daily 10-Minute IT Services Insights:
💬 https://www.podmatch.com/hostdetailpreview/businessoftech
🔗 Follow Business of Tech
LinkedIn: https://www.linkedin.com/company/28908079
YouTube: https://youtube.com/mspradio
Bluesky: https://bsky.app/profile/businessof.tech
Instagram: https://www.instagram.com/mspradio
TikTok: https://www.tiktok.com/@businessoftech
Facebook: https://www.facebook.com/mspradionews
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
[00:00:02] It's Wednesday, November 13th, 2024, and I'm Dave Sobel. Four things to know today. OpenAI and the AI industry pivot as scaling plateaus. AI's channel opportunity grows, but skepticism toward AI-capable PCs highlights an industry gap. CrowdStrike, Microsoft, and Box drive AI and security enhancements of rising enterprise demand. And Amazon's power request denied. This is the business of tech.
[00:00:30] In a recent report by Reuters, AI companies like OpenAI are facing significant challenges in developing new large language models due to unexpected delays and limitations in traditional scaling methods. AI researchers are now advocating for alternative training techniques that mimic human-like reasoning, which could reshape the competitive landscape of AI development. Notably, Ilya Sützkeber, co-founder of OpenAI, highlighted that the era of simply scaling
[00:00:59] scaling up models has plateaued, indicating a shift toward innovative approaches. OpenAI's latest model, known as O1, utilizes a technique called test-time compute, allowing models to evaluate multiple possibilities in real-time, which could enhance performance significantly. This transition poses implications for AI hardware demand, particularly for NVIDIA's chips, as the industry moves towards cloud-based inference systems. According to Sequoia Capital, this shift could
[00:01:29] fundamentally change how AI models are developed and deploy. OpenAI is adjusting its strategy as the pace of improvements to its GPT-AI models begins to slow. This shift comes amid growing competition in the AI landscape and a need to focus on sustainable innovation. Recent data indicates that while user engagement remains high, the rate of new feature deployment has decreased significantly since last year.
[00:01:56] A survey by the information revealed that 70% of AI researchers believe the current advancements are not meeting the expectations of previous breakthroughs.
[00:02:06] An OpenAI's latest model, Orion, has sparked discussions around the potential stagnation of those advancements as experts raise concern about hitting that scaling wall. Industry analysts suggest that the future of AI will depend on breakthroughs in hardware and algorithm development.
[00:02:21] An OpenAI's chat GPT continues to lead the AI landscape, with monthly visits soaring to nearly 4 billion, marking a 100% increase from 2023, according to SimilarWeb.
[00:02:33] In October, the platform received a 17.2% rise in users compared to September, achieving 3.7 billion visits.
[00:02:42] Impressively, chat GPT now surpasses Google Chrome in user numbers, reaching this milestone in just two years, compared to Chrome's 16%.
[00:02:51] Meanwhile, Google's Notebook.com, an AI note-taking service launched in July 2023, has seen a growth surge of 300% in September and 201% in October.
[00:03:04] Other AI tools like Microsoft's Copilot and Claude AI have also reported significant traffic increases, with Copilot reaching 64.9 million visits in October, a rise of 87.6%.
[00:03:18] Why do we care?
[00:03:20] The acknowledgement by AI researchers that scaling alone has hit diminishing returns marks a crucial inflection point.
[00:03:28] Moving forward, competitive advantage will increasingly hinge on differentiated training techniques, which could reshuffle market leadership and could have underlying models become commodities.
[00:03:39] The survey, indicating that 70% of researchers feel current advancements aren't meeting the expectations, reflects a broader sentiment that AI innovation is entering a period of recalibration.
[00:03:51] We're right in the trough of disillusionment as practical applications start to appear.
[00:03:57] As the industry hits that scaling wall, the focus will likely shift toward efficiency, cloud-based inference, and algorithmic breakthroughs that mimic human reasoning, ultimately fostering a competitive, differentiated market.
[00:04:11] For tech providers, the implications are substantial.
[00:04:13] They must not only adapt to hardware change demand and model development strategies, but also anticipate longer, less predictable cycles of innovation.
[00:04:25] Thread has declared death to the ticket.
[00:04:28] Did you know tickets date back to the 1800s?
[00:04:31] And customers hate tickets.
[00:04:34] Thread uses an approach for connecting, communicating, and ultimately collaborating.
[00:04:38] Thread allows people to come together around a topic.
[00:04:42] They can discuss and make decisions.
[00:04:44] They can share and invite the right people.
[00:04:47] The future of service is collaborative.
[00:04:50] Supercharge your service experience by seamlessly integrating AI and automation to meet your customers where they are.
[00:04:56] In Teams, Slack, and desktops.
[00:04:59] Not in tickets.
[00:05:00] With instant updates to and from ConnectWise, Autotask, and Halo PSA, keep all communication in one place, where it should be.
[00:05:09] With people.
[00:05:10] Decrease time to resolution 30% with chat-based support by visiting getthread.com slash MSP radio to declare death to the ticket.
[00:05:22] New research from Canalys reveals that while optimism about AI is growing in the channel, enthusiasm for AI-capable PCs is lagging.
[00:05:31] Despite a 49% growth in AI PC sales last quarter, these devices only account for 20% of the market.
[00:05:39] Over 40% of channel partners see AI as a significant growth opportunity.
[00:05:44] Yet more than a third do not plan to sell Copilot Plus PCs next year.
[00:05:48] A quarter of respondents noted that announcements regarding AI PCs have influenced customer refresh plans.
[00:05:55] But 15% reported delays in purchasing until these devices become available.
[00:06:01] Canalys stresses the need for clearer value propositions and better education for partners to effectively promote AI-capable PCs,
[00:06:10] especially as Windows 10 support ends soon.
[00:06:12] User resistance remains high due to security and ethical concerns necessitating transparency and collaboration among the industry stakeholders.
[00:06:22] A recent study by TeamViewer reveals that over 60% of executives trust AI to make decisions without human oversight,
[00:06:30] while 70% believe it can accurately forecast future business scenarios.
[00:06:34] In contrast, only half of IT decision makers share this trust,
[00:06:39] with just 30% in the UK comfortable allowing AI to make decisions autonomously.
[00:06:45] Despite concerns, three-quarters of IT professionals appreciate AI for enabling them to focus on strategic tasks and acquire new skills.
[00:06:54] Additionally, two-thirds of UK decision makers anticipate a positive impact on revenue due to AI,
[00:07:00] with C-level executives predicting an average increase of 270%.
[00:07:04] The research indicates a growing reliance on AI,
[00:07:08] with 82% of UK decision makers engaging with the technology weekly,
[00:07:12] a significant rise from last year.
[00:07:14] A recent Slack report reveals that the rate of AI adoption among desk workers is slowing,
[00:07:21] with a mere three-percentage-point increase globally over the last three months.
[00:07:26] This follows a period of rapid growth where adoption surged from 20% to 32% in the preceding six months.
[00:07:33] The survey, which included over 17,000 employees and was conducted by Qualtrics,
[00:07:39] highlights that nearly half of desk workers feel the need to hide their AI usage from managers,
[00:07:44] fearing perceptions of incompetence.
[00:07:47] Despite these mixed signals, 97% of executives express a strong urgency to incorporate generative AI into their operations next year.
[00:07:57] Leaders remain optimistic about overcoming AI-related challenges,
[00:08:02] with two-thirds confident in their organization's risk management capabilities.
[00:08:06] As the hype surrounding AI diminishes,
[00:08:09] experts believe there's still significant potential for AI to enhance job satisfaction and efficiency.
[00:08:16] Why do we care?
[00:08:18] Well, AI's long-term success in business hinges not only on those advances,
[00:08:23] but on bridging trust gaps,
[00:08:25] aligning tools with tangible workplace benefits,
[00:08:28] and establishing a supportive environment that demystifies AI's role in operations.
[00:08:33] The AI PC market's growth can't be sustained by novelty alone.
[00:08:38] Experts note that as initial hype fades,
[00:08:41] there's room for AI to settle into its role as a tool that enhances efficiency.
[00:08:46] As the industry moves into a phase of normalization,
[00:08:50] clear, transparent communication from leaders and vendors
[00:08:53] on the practical benefits and limitations of AI tools
[00:08:56] will help refine expectations and sustain long-term adoption.
[00:09:03] CrowdStrike has launched its AI Red Team services,
[00:09:06] aiming to enhance the security of AI systems against rising cyber threats.
[00:09:11] As organizations rapidly adopt artificial intelligence,
[00:09:14] risks such as model tampering and data poisoning are increasing.
[00:09:18] The new service was announced at CrowdStrike's premier user conference,
[00:09:22] fal.com Europe,
[00:09:24] providing comprehensive security assessments to identify vulnerabilities in AI systems,
[00:09:28] particularly large language models.
[00:09:31] Key features include proactive defense mechanisms aligned with
[00:09:34] OWASP top 10 LLM attack techniques
[00:09:39] and tailored attack scenarios to test systems against relevant threats.
[00:09:43] Microsoft has acknowledged that some instances of Windows Server 2019 and 2022
[00:09:48] were unexpectedly upgraded to Windows Server 2025
[00:09:52] due to mismanagement by third-party update tools.
[00:09:55] The company has confirmed the issue via its release health dashboard,
[00:10:00] advising users to check their update configurations.
[00:10:03] While the Server 2025 feature update was marked as an optional release,
[00:10:09] some users experienced surprise installations,
[00:10:12] with reports indicating that even standalone servers were affected.
[00:10:16] Cybersecurity firm Heimadal highlighted that Microsoft
[00:10:19] had misclassified the update before it was retracted.
[00:10:23] Although the problem has been labeled as mitigated,
[00:10:26] Microsoft has not yet provided rollback tools,
[00:10:29] leaving administrators to rely on backups
[00:10:31] or potentially incur licensing costs for the new server version.
[00:10:35] The company is collaborating with third-party providers
[00:10:37] to enhance update procedures.
[00:10:39] And Microsoft is rolling out a significant update for Windows on ARM,
[00:10:43] enabling more X64 software and games to run through its Prism emulation.
[00:10:48] The update, part of the Windows 11 Insider Preview build 27744,
[00:10:54] is designed to work with Qualcomm's Snapdragon X Elite and X Plus processors.
[00:11:00] The enhancement allows these chips to run applications that lack native ARM 64 versions,
[00:11:06] including popular titles like Starfield and Helldrivers 2,
[00:11:10] which utilize process extensions like AVX and AVX2.
[00:11:14] While many applications such as Photoshop and Chrome already have ARM 64 versions,
[00:11:20] some still require emulation and may experience limitations.
[00:11:24] The emulator is currently in limited use on retail PCs,
[00:11:28] and while it supports 64-bit x86 software,
[00:11:32] 32-bit applications remain incompatible.
[00:11:35] And Box continues to evolve from a file-sharing platform
[00:11:39] into a comprehensive enterprise AI provider
[00:11:41] with the introduction of Box AI Studio and Box Apps.
[00:11:45] Announced at the BoxWorks event,
[00:11:48] Box AI Studio enables organizations to create custom AI agents
[00:11:52] tailored to their specific needs,
[00:11:54] built on partnerships with players like Anthropic, Google, and OpenAI.
[00:11:59] CEO Aaron Levy highlighted that this marks a transformative moment,
[00:12:03] aiming for a revenue growth pathway from $1 billion to $2 billion
[00:12:07] by enhancing content management capabilities.
[00:12:10] Additionally, Box Apps allows enterprises to develop
[00:12:13] no-code applications directly within the Box environment,
[00:12:17] streamlining workflows such as contract management
[00:12:19] and invoice processing without the need for extensive customer development.
[00:12:25] Why do we care?
[00:12:26] Well, CrowdStrike's AI-focused security underscores the need
[00:12:29] for tailored cybersecurity solutions as AI applications proliferate.
[00:12:34] As more enterprises incorporate AI,
[00:12:37] understanding and defending against AI security threats
[00:12:40] will become table stakes for cyber firms
[00:12:42] and a critical focus for enterprise risk management.
[00:12:45] The Windows on ARM updates extends ARM's reach of the enterprise,
[00:12:49] particularly as ARM's efficiency gains
[00:12:51] attract interests from organizations
[00:12:53] focused on sustainability and cost reduction.
[00:12:56] And I'm watching this closely as we track Intel's decline.
[00:13:02] The Federal Energy Regulatory Commission has denied Amazon's request
[00:13:07] for an increased power supply from the Susquehanna nuclear plant,
[00:13:11] which aimed to boost capacity from 300 to 480 megawatts
[00:13:15] to support its expanding data center operations.
[00:13:18] Regulators expressed concern that the approval
[00:13:20] could jeopardize grid reliability and escalate energy costs,
[00:13:24] as demand for data center power driven by AI workloads
[00:13:28] is projected to rise by 160% by 2030,
[00:13:32] according to Goldman Sachs.
[00:13:34] Tech firms are increasingly considering
[00:13:36] co-locating data centers near nuclear facilities.
[00:13:40] Meanwhile, companies like Constellation Energy and Microsoft
[00:13:42] have formed partnerships to revive dormant nuclear sites,
[00:13:46] reflecting the ongoing challenges in securing reliable energy
[00:13:49] for tech operations.
[00:13:52] Why do we care?
[00:13:53] Well, partnerships with nuclear providers or other energy sources,
[00:13:56] while complex, could offer stability in a landscape
[00:13:59] where grid reliance is becoming unsustainable.
[00:14:02] Expect an increase in partnerships with energy providers,
[00:14:06] particularly those in nuclear and renewable sectors,
[00:14:09] as companies look to secure power
[00:14:11] while aligning with their sustainability commitments.
[00:14:14] As regulatory and cost pressures mount,
[00:14:16] data centers will be pressed to maximize energy efficiency.
[00:14:19] This could drive innovation in cooling, server optimization,
[00:14:23] and AI-driven energy management systems,
[00:14:26] making energy efficiency a differentiator for data center operations.
[00:14:30] And that all has a number of second-order effects worth tracking.
[00:14:37] Today's episode is supported by CoreView.
[00:14:40] Your customers need your Microsoft 365 expertise,
[00:14:43] and CoreView has the only M365 management platform designed for MSPs.
[00:14:49] CoreView has the only M365 management platform designed for MSPs.
[00:14:49] Manage hundreds of tenants, automate manual tasks,
[00:14:52] and monitor compliance,
[00:14:54] all while intelligently comparing to the baseline.
[00:14:56] With a no-code control approach,
[00:14:59] CoreView revolutionizes your Microsoft 365 administration.
[00:15:03] This powerful platform enables automatic reporting and remediation,
[00:15:07] ensuring optimal performance and security.
[00:15:09] The best part?
[00:15:11] You achieve this high level of service
[00:15:13] without the need for a large workforce,
[00:15:15] allowing you to focus on growing your business through efficiency.
[00:15:19] Want to know more?
[00:15:20] Visit coreview.com slash MSP and find out more.
[00:15:27] Thanks for listening.
[00:15:28] Today is National Dream Destination Day,
[00:15:31] National Huggy Musician Day,
[00:15:32] and World Kindness Day.
[00:15:34] All in all, a pretty positive day.
[00:15:37] Got a comment or a thought on a story?
[00:15:39] Put it in the comments if you're on YouTube,
[00:15:40] or reach out to me on LinkedIn if you're listening to the podcast.
[00:15:44] If you've enjoyed the show,
[00:15:45] give it a review,
[00:15:46] and make sure you've subscribed or followed on your favorite platform.
[00:15:49] I'll talk to you again tomorrow.
[00:15:53] The Business of Tech is written and produced by me,
[00:15:56] Dave Sobel,
[00:15:57] under ethics guidelines,
[00:15:59] posted at businessof.tech.
[00:16:01] If you like the content,
[00:16:03] please make sure to hit that like button,
[00:16:05] follow or subscribe.
[00:16:06] It's free and easy,
[00:16:08] and the best way to support the show
[00:16:10] and help us grow.
[00:16:11] You can also check out our Patreon,
[00:16:14] where you can join the Business of Tech community
[00:16:16] at patreon.com slash MSP radio,
[00:16:19] or buy our Why Do We Care merch
[00:16:22] at businessof.tech.
[00:16:24] Finally, if you're interested in advertising on this show,
[00:16:27] visit MSP radio.com slash engage.
[00:16:31] Once again,
[00:16:32] thanks for listening to me,
[00:16:33] and I'll talk to you again
[00:16:35] on our next episode
[00:16:36] of the Business of Tech.
[00:16:40] Part of the MSP Radio Network.

