Signal over Noise with Ulrike Hoffmann-Burchardi

Hello and welcome to Signal over Noise. I'm Ulrika Hofmann-Borchati, CEO for the Americas and Global Head of Equities for UBS Wealth Management. AI is tech, but tech is not AI, was the clearest and most important market signal last week, maybe even for the year.

AI is redrawing the lines between winners and losers, and that includes the technology sector itself. The software ETF IGV was down 9% on the week, and now down 24% on the year. The reason was Anthropic's release of legal and financial tools as part of its co-work offering.

And these new releases happened quickly. Anthropic's tools went from concept to products in months, not years. AI will likely yield the most profound productivity boom in human history, the extent of which we still likely underestimate.

But we likely also underestimate AI's disruptive nature, even if it pales against its benefits. AI will make any digital product or service more intelligent, from application software to advertising. AI is not more than an algorithm, it is software itself.

And one of its superpowers is writing code. The productivity increases from AI developer tools are not new. Studies have shown 50% plus productivity increases for developers.

But what the market under-appreciated is how quickly the AI agents may replace existing software. Our view has long been that existing software will be disrupted by AI, and we continue to believe this is the case. And this is for two reasons.

One, software written and maintained by AI will be much less costly, even if AI just replicates existing features. And two, traditional software is a workflow tool and not inherently intelligent. AI agents can complete tasks, they can access databases, and get to the desired outcomes by themselves.

So software built by AI and leveraging AI in its functionality will be both much cheaper and much more performant. Software-as-a-service companies own the interface with dashboards, but AI agents own the outcome. If AI can replicate the outcome, the software-as-a-service layer becomes redundant.

Software replacements in the enterprise are not straightforward though. The move from on-premise software to software-as-a-service took 15 to 20 years from inception to mainstream dominance. It started with small and mid-sized companies adopting SaaS in the early 2000s, before cloud infrastructure led to large-scale deployment in the enterprise post-2010.

For AI, the limiting factor for agentic AI will be purely organizational. How quickly will enterprises allow the adoption of agentic AI at scale? It won't be immediate, mainly because of data security and auditability concerns, but a lot faster than the SaaS adoption, because the return from deploying these AI tools will be magnitudes larger.

So in the short term, the software sell-off in the public markets is likely overdone, but the long-term prospects of traditional software remain under pressure. We continue to advise to be highly selective with long-duration commitments to traditional software in both private and public markets. Last week, we also received Amazon's AI Data Center CapEx figure, completing the current picture for the four US hyperscalers.

A combined 600 to 620 billion spend for 2026. This number exceeded analysts' expectations by about 130 to 150 billion, mainly driven by Google and Amazon. 600 billion of annual CapEx is a staggering number. Investors have likely become immune to these large CapEx numbers.

But to put it into perspective, this CapEx amounts to 2% of US GDP. And it amounts to the CapEx that all of the four hyperscalers spent over a total of 15 years to build out cloud computing from 2006 to 2021. This number clearly bodes well for compute demand, but it does not bode well for the cash flows and profit margins of the big four.

In fact, their free cash flow will likely be down from more than 200 billion in 2024 to negative in 2026, accompanied by margin erosion. This will put pressure on private AI companies to increase CapEx to remain competitive. For private companies, this is more difficult as they don't have the same amount of cash or cash flows to draw from.

The willingness of investors to support the cash needs will be one of the key metrics to watch, especially if private AI company margins will also come under pressure. Turning to the macro side, ISM on Monday was the first strong data point that the US economy is in a cyclical upswing. The ISM manufacturing PMI read 52.6, the highest reading since August 2022, and the first green shoot supporting a manufacturing recovery in the US.

We continue to believe that the fiscal and monetary stimulus in the US provides support for a cyclical recovery this year, in particular in the first part. Next week, we will finally get the delayed January non-farm payroll number and January CPI. CPI might show a seasonal uptick in inflation, as many companies revisit their pricing models at the beginning of the year.

We would not read too much into this as we continue to believe that lower wage and shelter inflation, together with a favorable base effect from tariffs this year, will lead to a more benign inflation outlook. All of this makes us optimistic about economic growth and a broadening of the equity market rally in 2026. With this, stay well and stay ahead. and brokerage services.

Investment advisory services and brokerage services are separate and distinct, differ in material ways, and are governed by different laws and separate arrangements. In the USA, UBS Financial Services Inc. is a subsidiary of UBS AG and a member of FINRA SIPC. For information, please visit our website at ubs.com forward slash working with us.

For a full legal disclaimer applicable to the independent investment views produced by UBS, please visit our website at ubs.com forward slash CIO dash disclaimer.