AI’s High-Stakes Game: The Risks and Opportunities Shaping Its Growth

If Generative AI is comparable to previous significant technological breakthroughs (such as personal computers, the internet, and smartphones), then some of the most promising investments might still be on the horizon, or they may be early-stage startups that aren’t widely recognized.

Thus, there’s no immediate need to invest in today’s dominant players; however, it’s beneficial to understand the types of companies that are worth monitoring.

The Generative AI Ecosystem

The expectations surrounding leading AI companies are undeniably elevated, which increases the likelihood that actual returns may fall short of these lofty goals. Suppose you’re looking to expand your investments in AI. In that case, one strategy to mitigate this threat is aimed at focus on lower value chain companies where expectations are less inflated.

McKinsey & Firm has outlined six key layers in the evolving generative AI ecosystem.

✨ Although the majority of the focus is currently on the foundational three layers, McKinsey anticipates that the greatest long-term value will accumulate at the upper tiers.

This might not be immediately apparent, but history suggests a similar trajectory. The initial winners of the web included infrastructure providers like Cisco—yet, over time, it was innovative firms as such as AWS and Alphabet that unlocked the most value.

Additionally, the lower layers demand massive capital investments, which is why they are largely controlled by well-funded industry giants. As the sector matures, there’s hope for broader participation from emerging players.

Now, let’s explore the various prospects and possible drawbacks across these six layers.

1. Computing Hardware

LLM (large languages model) demand substantial processing power, with GPUs at the heart of this need. However, data centers rely on various additional infrastructure and connectivity equipment.

Potential Advantages

Nvidia dominates the GPU market, placing it at the forefront of AI boom. AMD plays the challenger role, while TSMC is key manufacturer behind most GPUs. Other players in this sector include companies that provide servers, networking gear, and essential hardware for powering AI models. Notable examples are Supermicro, Arista Networks,…

Other players in semiconductor industry include ASML, Cadence, Synopsys.

Risks

While it is almost certain that Nvidia is poised to ship a significant number for GPUs in a coming years, there is a risk that its revenues or profit margins may need to meet the high expectations reflected in its share price. The lack of unique intellectual property (IP) to safeguard margins poses a potential risk for other hardware suppliers to data centers. Although constructing server infrastructure and networking gear is straightforward, is relatively straightforward, GPU design and manufacturing are more complex.

2. The Cloud Platforms

Major technology firms that provide computing capacity and data storage have capitalized on most innovations over the past decade and remain central to the AI landscape.

Opportunities

Leading cloud providers numerous advantages in harnessing AI, including the funding and infrastructure needed to build models, the ability to provide computing power and services, and a vast existing customer base for new AI offerings. Some companies specialize in developing and owning physical infrastructure. Notable examples consist of REITs such as Equinix, Digital Realty Trust,…. Additionally, private equity powerhouse Blackstone is committing $25 billion to a network of AI-focused data centers.

Risks

Although major tech firm are well diversified, they are only partially shielded from the possibility of AI losing its appeal. For other firms, including REITs, there is a constant risk of being driven by hype over solid fundamentals, making it crucial to assess valuations critically.

3. Foundation Models

Core generative AI model, such as GPT, Llama, Gemini are extensive pre-trained models that utilize text input to generate various outputs, compassing text, video, audio…. Applications such as ChatGPT utilize these frameworks as a foundational engine for producing specific content.

Opportunities

Developing and training LLM demands expertise, access to powerful computing resources, and extensive datasets. Consequently, this area is primarily dominated by large corporations, alongside startups such as Anthropic, OpenAI, which collaborate with them. Major players like Microsoft, Meta,… are leading in this space, though there are potential challengers as such as Apple and X that might emerge. Over time, as chip technology improves, smaller firms may find it easier to compete.

Risks

As previously noted, LLMs face challenges in maintaining a competitive edge, the rise of community-driven models.

4. Model repositories and Machine Learning Operation

Model repositories are platforms or services that store and organize models, frequently granting access to those held by cloud providers. MLOps, short for MLOs, encompasses the lifecycle of creating, deploying, and overseeing ML applications.

This value chain segment may be easy to diseregard, but it is crucial. Building AI applications entails significant costs and requires thorough design, evaluation, and oversight. The rapid evolution of technology further emphasizes the importance of these processes.

Opportunities

Major cloud providers all offer various MLOps services, there are many other companies offering comparable solutions.

These include established IT firms like IBM, Kyndryl, and HPE, along with artificial Intelligence and data-centric platforms such as Palantir, Snowflake, C3 AI. It’s probable that data-driven platforms will keep enhancing their AI capabilities in the future.

Additionally, a rising number of startups are joining the MLOps landscape, This list also features Databricks (speculated to be heading toward an IPO), Hugging Face.

Risks

When investing in this sector, it’s essential to comprehend the business models of these companies and the extent of their recurring revenue. While consulting firms typically require minimal capital, which can be advantageous, they may need more predictability of businesses based on subscription models.

5. Uses

Sometimes, LLMs feel like rough drafts—capable of producing impressive-looking content but often riddled with inaccuracies. That might not be a problem for creative tasks like storytelling, music composition, or video generation. But when precision matters, these models need refinement, along with continuous updates to stay relevant.

✨ Think of foundation models as the engine of a car. The applications they support? That’s everything else—transmission, brakes, wheels. These may differ based on the intended purpose: These can vary depending on the intended use: is it designed for speed or utility?

Generative AI applications are transforming industries, but to be useful, they need more than just raw model output. A well-designed interface, customization, and access to the right information make all the difference.

Take a bank’s AI-powered customer service assistant. Simply chatting with customers isn’t enough—it needs to retrieve account details, recall past interactions, and enforce strict privacy rules. Fine-tuning, high-quality training data, and well-structured prompts are what turn it from a generic chatbot into a valuable tool.

Chances

Anybody can build an AI-powered application, but not everyone has the same level playing field.

Larger companies have key advantages that small startups and individuals often lack—access to capital, top-tier expertise, and a massive user base. More importantly, they own vast amounts availability of extensive exclusive data, giving them a significant edge in training and fine-tuning AI models.

Tech giants like Meta, Microsoft, Amazon… check all these boxes. Other major players, including Salesforce, Shopify, Tencent,… and Alibaba, also hold strong positions in the AI race.

For the majority of these firms, AI potential is already factored into their valuations. However, one exception might be Apple. Analysts currently project modest revenue and earnings growth of around 5%, but that outlook could shift if Apple enters the AI space with groundbreaking tools or services.

Some analysts see Apple as the unexpected contender in the AI competition. Firms is developing its own large language model, and speculation suggests Siri will soon integrate generative AI. More importantly, with 2 billion units in use and a thriving App Store, Apple stands to benefit not just from its own AI advancements but also through third-party applications. While details remain uncertain, this is a development worth keeping an eye on.

Risks

The entry barriers for AI app development are relatively low. If history is any guide, a speculative frenzy—similar to what we saw with dotcom, EV, and cannabis stocks—could emerge, particularly among small-cap companies.

✨ Before investing, focus on businesses with real distinct strengths and a clear, sustainable growth trajectory.

🛠️ AI Product

At the top of McKinsey’s Generative AI value chain are the Services are companies that specialize in helping businesses build and implement AI solutions. A lot of these firms overlap with those offering MLOps services, providing essential tools and expertise for AI deployment.

🔍 AI Investment Opportunities

Mizuho Securities analysts have identified IT services firms as a less direct but promising way to gain exposure to AI. These firms have long been guiding businesses through digital transformation, giving them an established network and deep industry expertise.

IBM, Kyndryl, and HPE are key players in this field. Other companies highlighted include accenture, EPAM Systems, Globant. These firms are generally categorized within the IT consulting sector, although some may belong to different sectors.

The chart below from Simply Wall St’s Markets Page showcases P/E and growth forecasts for IT Consulting, the broader tech industry, and the overall market—offering a valuable tool for comparing industry expectations and potential upside.

Risks

Although these firms are involved in AI, they are not the key players driving the industry forward, so they will likely remain on the sidelines. This positioning can be beneficial because if AI hype collapses, the share prices of these firms are unlikely to experience a dramatic decline.

However, this also means that you are unlikely to uncover a rare 10-bagger investment in this sector, since their contributions are incremental, and their role is more supportive than transformative.

The Takeaway: Don’t Overlook Firm AI Adopters

Considering the uses layer of the AI value chain, we often focus on companies developing tools and services for consumers and businesses. However, those corporations that implement these applications—often even creating their hold—can reap significant benefits.

There are a few key areas where they can gain:

Administration: Enhanced efficiency and reduced operational costs.

Product Development: Accelerated prototyping and faster development cycles.

Marketing: Tailored content generation and valuable insights.

Customer Service: Improved engagement at lower costs through chatbots.

Businesses can boost revenue by utilizing the good tools while simultaneously cutting costs. Firms with substantial administrative expenses stand to gain significantly by minimizing these costs. Additionally, firms that produce large volumes of data can utilize that information to extract valuable insights.

This advantage spans numerous industries but particularly shines in sectors like banking, insurance, and healthcare. Organizations outside the tech realm that harness AI may also be less vulnerable to the repercussions if the AI bubble bursts.