This article summarizes an episode of SaaStr AI’s video series featuring Jason Lemkin, CEO and founder of SaaStr.

Jason Lemkin, CEO and Founder of SaaStr/ Photo credit: SaaStr

Global software spending is expected to reach a record high in 2026, but for most companies, that growth is misleading. Jason Lemkin, CEO and founder of SaaStr, sees the market splitting into two. A few AI leaders are growing quickly, while everyone else competes for what’s left in a hidden budget crisis.

The great divide in startup growth

Headlines are filled with stories of AI startups seeing huge growth, getting big in months, not years. This makes the market look like it is booming. In reality, a small number of companies are growing fast, while most are struggling to grow at all.

Getting big has never been faster
Lemkin says, “It’s literally never been easier to scale to US$100 million quickly, but only for a select few. Glean has already gone from zero to US$100 [million in ARR]… this year. Replit and ElevenLabs went to US$200 million already this year.”

The downside of hyperspeed
The same forces that help a few companies grow fast also make competition tougher. New ideas can be copied almost overnight.

Lemkin notes, “There’s so much competition. Folks are getting cloned so quickly… A company I invested in here that is totally awesome AI, very disruptive, I think, launched about 60 days ago, has four clones already.”

Most startups with investors are failing
This fast-changing market is splitting into two. Most startups are struggling, while investors pour money into the few growing quickly.

Lemkin reports, “If you’re an investor, most startups are not doing great today. They’re not. But the ones that are really breaking out at this crazy level, the investors are just frothing at the gums to put more money in.”

Venture capital’s game of concentration

The market’s instability is worsened by a shift in venture capital. Investors are backing fewer top companies, leaving most startups with less funding and more risk.

The way investors give out money has changed

• About half of all venture funding now goes into just four large deals, leaving less for other companies.
• The total number of investments is far lower than in 2021.
• Investors prefer established AI companies, where bigger checks feel less risky and are easier to exit.

Investors are chasing established AI leaders
This has led to a shift where investors who once backed early-stage startups are pulling away.

Lemkin argues, “So many investors that I’ve known over the years, so many actually classic SaaS and B2B investors that used to do seed… they’re doing [deals with] Anthropic at US$80 billion… Why are they doing this? … That’s actually where the easy money is.”

Early-stage investing has become less attractive
Lemkin explains the math, “It’s less work to make 10 times more money. That’s just math today… It just seems so silly when you can just put the money into a late-stage unicorn AI company, and it quintuples in a year. One year. Why would you do anything else?”

The software budget mirage

This situation, where a few winners get all the investment, is also happening with customer spending. At first glance, the future looks good. Tech executives are preparing to spend more on software than ever before.

Tech executives plan to spend more on software than ever
Lemkin highlights, “Software spend is accelerating to a record level next year, especially in the enterprise. A record level. It is reaccelerating the Chief Information Officer’s budget… they’re predicting it’s going to reaccelerate to 15.2% for software.”

The spending increase is not for new vendors
The truth is that most of this new money is already claimed.

Lemkin breaks it down, “Half of this record spend is going to [software] price increases… and 30% of it is going into new AI tools that they’re bringing in. It doesn’t leave much left for all the rest.”

The market is shrinking for non-essential tools
Lemkin warns, “If you’re not in [the category of] existing vendors that are able to increase pricing and you’re not in a top AI initiative, the number of [new] vendors being added is shrinking or flat. It’s harder. Your job is harder.”

What it takes to break through

With tough competition for less available money, the standard for new products is higher. Adding an AI feature is no longer enough to stand out or justify a higher price.

AI products must offer huge value in one of three ways: 

• Automate employee tasks to reduce salary costs.
• Boost worker productivity so they can operate at a much higher level.
• Offer a 10x productivity gain that makes old ways of working outdated.

A simple co-pilot does not justify a higher price
An AI assistant that offers only small gains will not persuade executives to reallocate planned spending.

Lemkin says, “I’m not going to pay you two or three times as much because you have a cool copilot or a nice chat. That’s not enough. It’s not enough. AI features don’t count.”

The budget comes from replacing or augmenting headcount
The real upside for AI comes from the salary budget, not the software budget. The strongest products let companies rethink how work is done by reducing manual effort or significantly increasing employee productivity.

“It doesn’t matter unless you’re replacing a large part of DevOps humans with AI,” Lemkin explains. “It doesn’t count because there’s no extra budget… They bought [AI support software] to get rid of headcount, and that is going to accelerate next year.”

This Too Long; Didn’t Listen (TL;DL) summary was AI-generated and human-reviewed. Read all summaries here.