Big Bold Business Predictions for 2026

OpenAI Gets Government Funding

It's no secret that OpenAI has been spending money as if there were no tomorrow.

The company has taken the term blitzscaling to an absurd extreme. With plans to spend up to $1.4 trillion on infrastructure over the next decade, it's hard to imagine how Sam Altman will actually fund it. He has already made commitments to spend $250 billion with Microsoft, $300 billion with Oracle, and recently signed a $38 billion deal with Amazon.

Beyond spending on compute infrastructure (because who doesn't want a data center in their backyard?), OpenAI is paying its staff more than any other tech startup in history. In 2025, OpenAI employees received an average of $1.5 million in stock-based compensation. These equity awards account for nearly 46% of the company's anticipated 2025 revenue.

While OpenAI is on track to generate an estimated $13 billion in revenue for 2025, the company is also expected to post a net loss of $8.5 billion, primarily due to the staggering costs of operating ChatGPT and Sora.

Meanwhile, OpenAI's investors are scrambling to keep the "money furnace" stoked. SoftBank, in particular, just closed a $40 billion investment deal with the company, keeping it afloat for a while longer.

Our prediction: in 2026, Sam Altman will successfully convince the Trump administration to allocate taxpayer funds to OpenAI. Last year, the president signed an executive order known as the Genesis Mission, which lays the groundwork for such a move. As we have noted previously, CEO Sam Altman and CFO Sara Friar have already floated the idea of the government backstopping loans for OpenAI's infrastructure buildout. Essentially, this would use taxpayer money to guarantee that lenders receive a return, even if OpenAI itself goes bankrupt.

Major Security Breaches due to AI-Coded Software

"Beware of bugs in the above code; I have only proved it correct, not tried it."
- Donald Knuth

Although public perception of AI agents soured in 2025 (assuming it was ever positive outside of "LinkedIn Lunatics"), the reality is that agent-driven software development made significant strides this past year.

While OpenAI and Gemini have targeted the consumer market, Anthropic has quietly become the leader in business-oriented AI tooling. 2025 saw the launch of Claude 4.5 Sonnet and Claude 4.5 Opus, which introduced "Computer Use" capabilities. This allows the model to navigate an IDE or a web browser and interact with developer tools just like a human. Claude 4.5 can handle 30+ hours of autonomous coding, maintaining coherence across massive codebases without human intervention.

In software engineering, writing tests to ensure correctness is standard practice. While LLMs are often criticized for hallucinations, these automated tests serve as built-in safeguards against blatant errors. This has allowed agentic AI to proliferate throughout the industry, for better or worse.

Our prediction for 2026 is at least one high-profile security breach caused by software written with agentic AI. Just because tests prove software performs its intended function does not mean it cannot be exploited. Most security bugs arise when the "happy path" works correctly, but attackers find ways to make the system behave in ways the creators never anticipated.

This risk will only be exacerbated by code written by agents that lack the full architectural context of the systems they are building.

Tesla Launches Robotaxi in Austin

Tesla initially announced Full Self-Driving (FSD) nine years ago, on October 19, 2016.

At the time, Elon Musk promised that all Tesla vehicles produced from that day forward would be equipped with the hardware necessary for "full self-driving capability."

Nearly a decade and many missed deadlines later, we may finally be on the verge of witnessing the first truly autonomous Tesla vehicles via the Tesla Robotaxi program.

Currently, about 30 Tesla vehicles are ferrying passengers around Austin, Texas. For months, these vehicles have required safety drivers, but the company has already secured permission to operate without them for training purposes. A shift to fully autonomous commercial operations appears imminent.

The internet remains deeply divided on this. While skepticism of Elon Musk is at an all-time high, Waymo has already proven that a fully autonomous taxi service is viable. Tesla "just" needs to prove their vision-only technology is up to the task.

Given that Tesla is indeed operating driverless vehicles, we expect the company to pass government regulation and start operating the fully autonomous Robotaxi service in Austin in 2026.

Housing Market Takes a Big Tumble

Housing prices surged during the zero-interest-rate environment of the pandemic. While rates were at rock bottom, aspiring entrepreneurs were snapping up inventory to flip or convert into Airbnb rentals.

Five years later, in the post-pandemic era, interest rates have remained relatively high, causing the housing market to stall. Prices haven't dropped to meet lower demand however, likely because sellers are unwilling to take a loss on their investments. They would rather wait it out.

That wait may be coming to an end. As 2026 approaches, the general consensus is that the Federal Reserve will continue to lower interest rates to support the economy. This should encourage homeowners to list their properties again.

However, mortgage rates are not perfectly correlated with the federal funds rate. Lenders are free to price in additional risk, and currently, lending feels precarious. Whether it's just "bad vibes" or genuine economic instability, lenders will price those risks into their loans.

The result? Sellers who have been holding back will flood the market, only to find buyers who still cannot afford current financing options. Sellers will be left with two choices: delist or slash prices.

Our prediction: in 2026, we will finally see a significant drop in housing prices.

Intel 18A Fails to Impress

Intel desperately needs its 18A process to be a success.

TSMC customers, unable to secure the wafers they need due to the insatiable demand from AI monsters like Nvidia, are desperate for an alternative. Samsung and GlobalFoundries have failed to deliver manufacturing nodes that compete with TSMC since the 5nm generation.

Intel’s only hope is to pull off a miracle with a high-performing 18A node, which would give potential Intel Foundry customers the confidence to commit to the upcoming 14A node.

Our prediction is that Intel will fail to deliver. The 18A node is already delayed and reportedly producing lower yields than required. While things may improve in 2026, "improvement" isn't enough. Intel needs a generational leap to survive, and that looks increasingly unlikely.

Most likely, TSMC customers will remain TSMC customers - paying a huge premium and hoping the AI-driven demand eventually dissipates.