Target has admitted something that a large number of companies have kept to themselves so far - AI has moved it from "nice to have" to infra tier, but the bill is getting pretty painful. According to India head Andrea Zimmerman, the chain has moved from a stage where it is "using AI somewhere" to a state where it is "fully running on AI", only that the change in pricing models at the big providers (OpenAI, Anthropic and co) has forced the company to put the brakes on and be much stricter about where it really lets AI go.
For investors, it's an interesting paradox: Target had three years of declining revenue under its belt, and new CEO Michael Fiddelke was planning to spend another $2 billion on stores, renovations, and AI - while at the same time the variable costs of the actual "brains" to deliver that turnaround were rising. So AI went from being a marketing buzzword to an item that architectural teams and senior management are addressing at the "what exactly does that do to our margins and capex" level.
From "we have AI somewhere" to "we're working with AI"
Zimmerman described two key shifts:
Target $TGT has moved from AI pilot projects to having AI embedded in core processes - inventory planning, pricing, personalization, promotions management, logistics.
But at the same time, it started pushing for "intentional use" - i.e. not AI everywhere, but AI where it has a clear business impact: faster inventory turnover, less depreciation, better product mix, higher basket per customer.
For the retail chain, this makes sense:
AI can tell you exactly what to stock in a particular location.
when to re-price seasonal items so they don't sell below cost
or where it makes sense to send your own brand into direct competition with a discounter
But every such "smart" feature now means specific token consumption and a bill to the model provider.
Token AI: why it started to hurt Target
The second level is purely cost. OpenAI, Anthropic and others are moving to a model: you pay for what you actually use - the number of tokens when you run models, the breadth of context, the type of model. This is for the large enterprise:
more flexible (you don't pay for capacity you don't use)
but more risky (dependence on the volume of requests that you miss if you don't have good governance)
Zimmerman said bluntly that this shift "forced Target to rethink strategy." So the AI debate is taking place on two levels:
in architectural forums (how to design systems to use tokens most efficiently)
at the senior leadership level (how many AI initiatives can we realistically afford this year without eating into the budget)
This is an important signal to investors: AI adoption is no longer binary (have / have not), but about whether the company also has financial leverage in hand - sensible cost management around models.
India as the backbone: 40% tech people, focus on analytics
Target India is no longer a backoffice but a hub where AI and data translates into reality.
Around 5,600 people work in Bengaluru, around 40% of Target's entire tech workforce.
The teams cover merchandising, digital, stores and supply chain.
The company is looking to further strengthen analytics there - especially the ability to turn huge volumes of data into quick insights: what's selling, where, at what price, how customers are responding to promotions, at what rate sentiment is changing.
The goal is to shorten the loop: data → insight → action. In retail, weeks make the difference: if we overprice a product when the customer has already left for a competitor, it's too late. AI and analytics in India are supposed to be exactly that "speed transfer".
Business context: three years of declining sales, $2 billion in AI and stores
Meanwhile, Target is not in a position to "painlessly experiment".
The company's revenue has been declining for the last three years - pressure on price, some customers shifting to cheaper alternatives, weaker demand for discretionary goods.
New CEO Michael Fiddelke has set a plan to increase capex by about $2 billion - a combination of new stores, remodels and AI initiatives.
So the company will increase investment when the top line is stagnant or declining - making it a classic "turnaround + tech" story.
From a stock perspective:
AI isn't just a cool topic for an earnings call
but a tool that Target needs to use to fight its way back into revenue and margin growth.
but at the same time, AI (via token pricing models) can make that plan more expensive
What should an investor address
Instead of asking "how much AI is Target using" it's more interesting to ask:
Where exactly is AI adding EBIT
Specific use-case: inventory management, pricing, promo personalization?
What are the expected impacts (lower markdowns, lower stock-outs, higher online conversions)?
How Target manages AI costs
Does it have internal metrics like "tokens/order" or "AI cost/saved markdown"?
Does it make more sense to build more on proprietary models vs. pure reliance on large providers?
Capacity to deliver change
40% tech people in India is an advantage on the cost side, but also an organizational test: how well can they handle remote alignment between US business teams and Indian engineering?