More Than 20 Times Apart: The AI Productivity Estimates Dividing Forecasters

Sonal Goyal

12 November 2026 05:49 PM AEDT

More Than 20 Times Apart: The AI Productivity Estimates Dividing Forecasters

Image Source: Shutterstock

You are reading a free article with opinions that may differ from the recommendation given by Kalkine in its paid research reports. Become a Kalkine member today to get access to our research reports, in-depth technical and fundamental research. Learn more

Start Your Free Trial Now!

Key Highlights

AI productivity estimates vary widely, and the range of forecasts is very large.
Data centre Investment in Australia has risen tenfold from a low base in three years.
US AI-related share prices have roughly quadrupled since the end of 2022.
Investment is flowing in on expectations of larger future gains.

The range of estimates for how much artificial intelligence could lift productivity over the next decade is so wide that the top of it is more than 20 times the bottom, and an RBA Assistant Governor says that uncertainty is the honest starting point.

Ask ten economists how much artificial intelligence will boost the economy and you may get ten different answers. Ask them to put numbers on it and the spread becomes striking. In a speech at the UBS Australasia Conference in Sydney on 9 November 2026, Sarah Hunter, Assistant Governor (Economic) at the Reserve Bank of Australia (RBA), noted that estimates of AI's productivity effect over the next decade range from "over 2 per cent per annum to just 0.1 per cent."

Divide the top of that range by the bottom and the high-end estimate is more than 20 times the low-end one.

Why the estimates diverge so widely

Productivity growth is a measure of how much output an economy produces for a given amount of input, such as hours worked. Small differences in productivity growth, compounded over a decade, translate into very large differences in living standards and in how fast an economy can grow without generating inflation.

Estimates of AI's impact differ because they rest on very different assumptions. Key unknowns include how quickly businesses adopt AI and whether gains in individual firms add up across the whole economy.

What firms are telling the RBA

Rather than rely solely on outside estimates, the RBA has drawn on its liaison program with businesses. The findings Ms Hunter presented show adoption moving quickly.

The share of firms with no AI adoption fell from about one-third in 2025 to almost zero in 2026. Over the same period:

Limited adoption rose from around 40 per cent to around 60 per cent
Moderate adoption rose from around 20 per cent to 30 per cent
Deep or transformative adoption rose from 8 per cent to 10 per cent

The productivity picture from those same firms is more mixed. Around 35 per cent reported no productivity gains from AI, around 30 per cent reported minor gains, 23 per cent moderate gains and around 11 per cent significant gains.

In other words, almost every firm in liaison is now using AI in some way, but only around one in nine reports significant productivity benefits. That gap between adoption and measurable payoff helps explain why economy-wide estimates vary so much. Ms Hunter summarised the aggregate evidence directly: "So far, there is limited evidence that AI has materially lifted aggregate productivity."

Hype, slop and genuine gains

Part of the difficulty in measuring AI's contribution is separating productive uses from unproductive ones. Ms Hunter observed: "The term 'AI slop' has entered our vernacular to capture unproductive uses." Activity that generates more content without generating more value can make AI look busy without making the economy more productive.

At the same time, investment is flowing in on the expectation of larger gains. Data centre investment in Australia has risen "10-fold (from a very low base) over the past three years," according to Ms Hunter. In the United States, AI-related company share prices have "roughly quadrupled since the end of 2022."

Those market valuations implicitly lean towards the more optimistic end of the productivity range. Ms Hunter noted that "around 5 per cent of Australian households' financial Wealth is invested in the Equity of AI-related companies, both directly and through superannuation." If the returns disappoint, she said, such Assets "will be exposed to sharp repricing and possibly disorderly adjustments". The width of the productivity range is therefore not just an academic debate: it bears on Financial Risk too.

Why the range matters for the RBA

For a central bank, productivity is closely tied to how fast the economy can grow without stoking inflation. The November 2026 Statement on Monetary Policy notes that GDP growth "has slowed this year to a little below its potential growth rate". If AI lifted productivity towards the top of the estimated range, that potential rate would rise. If it delivered only 0.1 per cent a year, the effect on potential growth would be marginal.

The RBA's current forecasts show GDP growth of 1.6 per cent over the year to December 2026, easing to 1.3 per cent by December 2027. Headline Inflation is forecast to peak at 4.5 per cent in the December 2026 quarter, and both headline and trimmed mean inflation are expected to remain above 3 per cent until the end of 2027.

Ms Hunter's view on inflation was that "Consistent with previous general-purpose technologies, AI's long-run effect on inflation is likely to be broadly neutral." So even if AI ends up near the optimistic end of the productivity range, it would not necessarily change the long-run inflation picture.

The workforce dimension

The productivity debate also intersects with jobs. Citing Jobs and Skills Australia, Ms Hunter said about 4 per cent of workers are highly exposed to automation from AI, while 31 per cent are highly exposed to AI augmenting their work. "So, exposure to AI does not necessarily mean Job loss," she said.

Liaison results on headcount were consistent with gradual change: about three-quarters of firms reported no change, around 12 per cent slower hiring or not filling roles, around 8 per cent reducing headcount and around 6 per cent increasing it. Her broader characterisation was that "AI will automate some tasks, complement workers in others and create new roles as the economy adapts."

The bottom line

Estimates of AI's contribution to productivity over the next decade span from more than 2 per cent a year to just 0.1 per cent, a gap of more than 20 times. The RBA's liaison shows adoption has become almost universal among firms it speaks to, but only around 11 per cent report significant productivity gains, and Sarah Hunter says there is limited evidence so far of an aggregate lift. Until clearer evidence emerges, that uncertainty matters for growth forecasts, for asset valuations and for how much the economy can expand without adding to inflation.

Download Free Report – Explore 3 Stock Ideas & Industry Insights

Unlock 3 stock ideas and key industry insights in our free report. This information is general in nature and does not consider your personal objectives, financial situation, or needs. It is not financial advice.

All investments involve risk—consider independent advice before making any investment decisions.

View 3 Research Reports

Disclaimer:

This article (“Article”) has been prepared by Kalkine Pty Limited (ABN 34 154 808 312) (Australian financial services licence number 425376) (“Kalkine”) and its related bodies corporate who are authorised to provide general financial product advice. Kalkine.com.au and its associated pages are published by Kalkine.

Any information/advice provided in this article is general in nature and does not take into account your objectives, financial situation or needs. You should therefore consider whether the information is appropriate for your objectives, financial situation and needs before acting upon it.

There may be a Product Disclosure Statement, Information Memorandum or other offer document (“Offer Document”) for the securities or other financial products referred to in Kalkine articles. You should obtain a copy of the Offer Document and consider it before making any decision about whether to acquire the security or financial product.

Kalkine strongly recommends that you seek advice from a financial adviser, stockbroker or other professional (including taxation and legal advice) before acting on any advice/information in this Article or on the Kalkine website. Not all investments are appropriate for all people.

The information in this Article and on Kalkine website has been prepared from a wide variety of sources, which Kalkine, to the best of its knowledge and belief, considers accurate. Kalkine has made every effort to ensure the reliability of the information contained in its articles (including this Article), newsletters and websites. All information represents our views at the date of publication and may change without notice.

The information in this Article does not constitute an offer to sell securities or other financial products or a solicitation of an offer to buy securities or other financial products.

Kalkine does not issue, sell or deal in any financial products.

This Article may contain information on past performance of particular investments. Please note past performance is neither an indicator nor a guarantee of future performance.

To the extent permitted by law, and excluding any dishonesty or gross negligence by Kalkine, Kalkine disclaims and excludes all liability for any direct, indirect, implied, punitive, special, incidental or other consequential loss or damage arising from the use of or reliance on this Article, the Kalkine website and any information published on the Kalkine website without any warranties or representations by Kalkine to you. To the extent the law prohibits or limits this exclusion, Kalkine limits its liability to the resupply of services.

Please also read our Terms & Conditions and Financial Services Guide for further information.

Employees and/or associates of Kalkine and its related entities may hold interests in the securities or other financial products covered in this Article or on the Kalkine website. Any such employees and associates are required to comply with certain safeguards, procedures and disclosures as required by law.

Some of the images/music that may be used in the Article are copyright to their respective owner(s). Kalkine does not claim ownership of any of the pictures displayed/music used in the Article unless stated otherwise. The images/music that may be used in the Article are taken from various sources on the internet, including paid subscriptions or are believed to be in public domain. We have used reasonable efforts to accredit the source wherever it was indicated or was found to be necessary.

Kalkine Media Pty Ltd, an affiliate of Kalkine Pty Ltd, may have received, or be entitled to receive, financial consideration in connection with providing information about certain entity(s) covered on its website.