Earnings season is when companies reveal more than just how much money they made. The numbers matter, of course, but sometimes the most interesting clues come from what CEOs say about their plans, the challenges they face, and where they believe the next big opportunity lies.
The latest earnings calls from Tesla, Meta and IBM offered three very different perspectives on artificial intelligence (AI). Tesla is working to turn humanoid robots into useful machines, Meta is exploring how AI agents could change online shopping, and IBM is raising an important question about whether businesses are getting enough value from their AI investments.
All three companies see opportunities in AI, but each faces a different challenge. Tesla needs to make its robots more capable, Meta needs to turn growing interest in AI into a sustainable business, and IBM believes companies must pay closer attention to the returns they are getting from AI spending.
Here is what stood out from their latest earnings discussions and why it matters to investors.
1. Tesla: Why its Optimus robots need more time to learn
When people think of Tesla, electric cars usually come to mind. But Elon Musk has repeatedly highlighted the company’s ambitions beyond vehicles, particularly its humanoid robot project, Optimus.
The latest earnings discussion offered a glimpse into one of the biggest challenges facing that project: teaching robots to operate reliably in the real world.
Why training a robot is different from training a car
Tesla has an important advantage when developing its driving technology. Its vehicles generate large amounts of real-world driving data, which can help the company improve its software.
Optimus does not yet have the same advantage.
There is no comparable fleet of humanoid robots performing everyday tasks at scale and generating training data. Tesla therefore needs to produce robots that can practise tasks and learn through repeated attempts.
Musk compared the process to how a child learns to navigate the world. A toddler does not immediately know how to walk, read or complete everyday activities. These abilities develop through practice, experimentation and repetition.
His broader point was that Optimus needs to learn by doing, and it needs to do so at scale.
That sounds straightforward, but turning the idea into a commercially useful product is a much bigger challenge. A robot must do more than demonstrate a task in a controlled environment. It needs to perform consistently, handle different situations and operate safely around people.
What does this mean for Tesla’s spending?
Tesla also lowered its planned capital expenditure for the year to $20 billion from the $25 billion forecast three months earlier.
The discussion suggested that the additional training Optimus requires is one factor affecting the company’s plans.
For investors, this is worth watching. Developing a new technology requires significant investment, but the timing of that spending and the speed of commercial adoption can influence a company’s financial performance.
If Optimus eventually performs useful tasks reliably, it could open up opportunities beyond Tesla’s existing automotive business. However, the scale and timing of those opportunities remain uncertain.
What investors should watch
- Training progress: Can Tesla improve Optimus’s ability to perform tasks consistently in real-world environments?
- Commercial deployment: When will the robots be ready for practical use beyond testing and demonstrations?
- Capital expenditure: How much will Tesla need to spend before Optimus becomes a meaningful business?
- Financial returns: Can the company turn its investment in humanoid robots into a commercially viable product?
The big takeaway is that Tesla’s robotics ambitions remain closely tied to its ability to solve a difficult training problem. The opportunity may be significant, but investors still need evidence that the technology can move from development to practical use.
2. Meta: Could AI agents change how we shop online?
Meta’s earnings call highlighted another potential use for AI: personal assistants that help people complete tasks, including shopping.
The company’s Muse AI agent is already attracting interest from users. According to the earnings discussion, shopping has become its most common use case, with more than a quarter of users engaging in it and the average purchase exceeding $200.
Meta also said it had to work with Stripe to raise credit limits during the product’s first week because transaction volumes were higher than expected.
That is an interesting early signal. It suggests that users are exploring ways to use AI for activities that go beyond asking questions or generating content.
But it also raises a question that investors should not ignore: What happens when AI agents start making purchases on behalf of consumers?
From recommendations to transactions
Online shopping has traditionally required consumers to search for products, compare prices, read reviews and complete purchases themselves.
An AI agent could potentially take on some of that work.
Imagine asking an assistant to find a pair of running shoes within a certain budget, compare available options and help you choose the best fit. Instead of navigating multiple websites, you could rely on an AI tool to handle much of the process.
As these systems become more capable, they could change how consumers discover products and interact with retailers.
For Meta, this could create new commercial opportunities. The company already operates major social platforms where users discover products, interact with brands and encounter advertisements.
If its AI tools become part of the shopping process, Meta could potentially connect product discovery with purchasing in new ways.
However, the company still needs to demonstrate how this activity will translate into lasting revenue and profits.
The risk behind AI-powered shopping
The same development also creates questions around spending limits, payment authorisation and financial responsibility.
If an AI agent can initiate or facilitate purchases, users need clear controls over how much it can spend and what decisions it can make independently.
Retailers and payment providers also need reliable systems for handling transactions, disputes, refunds and potential fraud.
There is another concern: what happens if an AI agent makes an expensive purchase that the user did not intend to make?
These are not reasons to dismiss the technology. They are practical challenges that need to be addressed before AI-led shopping becomes a routine part of online commerce.
The early transaction activity reported by Meta indicates interest, but it does not yet establish how profitable this business could become.
Meta’s larger AI ambition
The shopping opportunity is part of a much broader investment strategy.
CEO Mark Zuckerberg has said Meta intends to build increasingly capable AI models by scaling up its technology. The company is developing larger models while also working on smaller, more efficient versions that can power personal assistants such as Muse.
The strategy is to improve AI capabilities while making those capabilities practical for everyday users.
Meta has also indicated that its infrastructure investments will remain substantial. It expects capital expenditure of $135 billion to $145 billion in 2026, according to the earnings coverage provided.
That is a major financial commitment, making the returns from its AI investments an important issue for shareholders.
What investors should watch
- User adoption: Will people continue using Muse after the initial excitement fades?
- Shopping activity: Can Meta turn AI-assisted purchases into a meaningful source of revenue?
- Spending controls: Can the company build trust around transactions initiated through AI agents?
- Infrastructure costs: Can improvements in AI efficiency help offset the cost of developing and operating these systems?
- Monetisation: Will AI generate enough additional business to justify the scale of Meta’s investment?
Meta’s opportunity goes beyond building another AI chatbot. It is trying to make AI part of how people use its platforms and carry out everyday tasks. The challenge is proving that this behaviour can support a sustainable and profitable business.
3. IBM: The question every company should ask about AI spending
While Tesla is focused on training robots and Meta is exploring AI-powered shopping, IBM’s earnings discussion highlighted a concern that applies across the corporate world.
Are businesses getting enough value from their AI investments?
IBM CEO Arvind Krishna said larger enterprises are becoming more aware of the true return on investment from AI tools. He also noted that some companies are concerned about spending getting out of control without delivering meaningful returns.
That is an important distinction.
There is a difference between adopting AI because it is attracting attention and adopting it because it solves a real business problem.
AI adoption is not the same as business value
Companies across industries are experimenting with AI to automate tasks, improve customer service, analyse information and help employees work more efficiently.
These applications can deliver real benefits. But the results depend on how the technology is implemented and whether it addresses a clearly defined need.
For example, an organisation might invest in an AI tool to reduce the time employees spend processing documents. If the tool works reliably and frees employees to focus on more valuable tasks, the investment could make sense.
But if the company spends heavily on the software, computing infrastructure and implementation without seeing meaningful improvements, the business case becomes harder to justify.
The same principle applies to more ambitious AI projects. A sophisticated model is not automatically a profitable one.
The real test is whether the technology delivers measurable improvements relative to its cost.
Why the return on investment matters now
AI development can require substantial spending on computing infrastructure, software, talent and ongoing operations.
Businesses therefore need to consider both the immediate costs and the longer-term benefits.
Some returns may be relatively easy to measure, such as lower processing costs or faster completion of routine tasks. Others, including better decision-making or improved customer experiences, may take longer to evaluate.
This can make it difficult to judge the success of an AI project in its early stages.
However, companies cannot rely on promises indefinitely. As spending grows, management teams and investors will want clearer evidence that these investments are improving productivity, supporting revenue or reducing costs.
IBM’s comments suggest that this scrutiny is becoming more important among large enterprises.
What this means for the wider technology sector
The shift towards measuring AI returns could influence how businesses choose their technology partners and allocate budgets.
Companies may become more selective about which tools they adopt, favouring solutions that address specific problems over projects that offer impressive demonstrations but unclear financial benefits.
They may also place greater emphasis on efficiency, reliability, security and integration with existing systems.
For technology providers, this creates both an opportunity and a challenge.
Businesses that can demonstrate tangible benefits may be better positioned to win long-term customers. Those that cannot explain the value of their products could find it harder to justify premium prices.
IBM’s message is not that AI lacks value. Krishna made clear that he believes AI can deliver meaningful productivity gains and change how work gets done.
The point is that businesses need to distinguish between AI’s genuine potential and the financial results they are actually achieving.
What investors should watch
- Enterprise demand: Are companies continuing to adopt AI tools, and which applications are attracting spending?
- Customer returns: Can technology providers demonstrate measurable productivity gains or cost savings?
- Spending discipline: Are businesses prioritising projects with clearer financial benefits?
- Profitability: Can AI companies grow revenue without allowing infrastructure and operating costs to rise just as quickly?
- Long-term contracts: Are customers willing to keep paying for AI solutions after the initial experimentation phase?
For investors, this makes IBM’s perspective particularly relevant. AI adoption may continue to expand, but the companies that benefit most could be those that help customers turn the technology into measurable business results.
4. Three companies, three different AI challenges
Tesla, Meta and IBM are approaching AI from different directions, but their earnings discussions point to a common theme: technological progress needs to translate into practical value.
Here is how their priorities compare.
| Company | Main AI focus | Key challenge | What investors should monitor |
|---|---|---|---|
| Tesla | Optimus humanoid robots | Training robots to perform real-world tasks reliably | Development progress, capital expenditure and commercial deployment |
| Meta | Muse AI agents and increasingly capable models | Turning user adoption into sustainable revenue | User engagement, shopping activity, infrastructure costs and monetisation |
| IBM | Enterprise AI solutions | Ensuring customers receive measurable returns on AI spending | Enterprise demand, customer outcomes and profitability |
Tesla is working towards a future in which robots can carry out useful physical tasks. Meta is trying to make AI a more active part of digital life, potentially including shopping and transactions. IBM is focused on the business case for AI and the value customers receive from implementing it.
The opportunities are different, but each company must show that its investment can support a viable business.
5. What should investors take away from these earnings calls?
One of the biggest mistakes investors can make during a technology boom is to focus only on the size of the opportunity.
AI could reshape industries, improve productivity and create entirely new products. But a large potential market does not guarantee that every company investing in it will earn attractive returns.
These earnings discussions offer three useful reminders.
First, innovation takes time. Tesla’s Optimus project illustrates how difficult it can be to move from a promising concept to a reliable product. Progress matters, but so do the costs and the time required to achieve it.
Second, adoption needs to lead somewhere. Meta’s early shopping activity suggests that users are exploring new ways to interact with AI. The next question is whether that activity can generate durable revenue without introducing unacceptable risks.
Third, spending must eventually produce results. IBM’s comments reflect a growing need for businesses to evaluate AI investments based on their actual impact, rather than the excitement surrounding the technology.
Investors should also remember that a strong earnings report does not automatically make a stock attractive at any price. Valuation, future growth expectations, competitive pressures and the amount a company needs to spend all influence its investment case.
The share price reaction on a particular day is only one part of the picture. What matters over the longer term is whether a company can execute its strategy and generate financial results that justify investors’ expectations.
The bottom line
The latest earnings calls from Tesla, Meta and IBM show three different stages of the AI opportunity.
Tesla is still working through the technical challenges of making humanoid robots useful at scale. Meta is exploring how AI agents could become part of everyday activities such as shopping. IBM is reminding businesses that AI spending needs to deliver real productivity gains and financial value.
None of these stories has a guaranteed outcome.
For investors, the important question is not simply which company is investing the most in AI or making the boldest promises. It is which company can turn its technology, spending and customer demand into sustainable business results.
AI’s potential is enormous, but the companies that execute well and manage their costs may ultimately matter more than the companies making the biggest claims.
What do you think? Which of these three companies is best positioned to turn its AI investments into lasting business value: Tesla, Meta or IBM?