Why is my plot of the cost function like this and not like a bowl?

My code:

def calc_cost_function(w, b, data):
    m = len(data)
    cost = 0
    for i in range(m):
        x = data.iloc[i].X
        y = data.iloc[i].Y
        cost += ((x * w + b) - y)**2
    return cost / (2 * m)


def gradient_descent(w_current, b_current, data, LR):
    w_gradient = 0
    b_gradient = 0
    m = len(data)

    for i in range(m):
        x = data.iloc[i].X
        y = data.iloc[i].Y
        w_gradient += (1 / m) * (x) * ((w_current * x + b_current) - y) 
        b_gradient += (1 / m) * ((w_current * x + b_current) - y) 
    
    w = w_current - LR * w_gradient
    b = b_current - LR * b_gradient
    
    return w, b

# Run gradient descent
w = 0
b = 0
LR = 0.0001
epochs = 1000
for i in range(epochs):
    w, b = gradient_descent(w, b, bigdata, LR)

print(f"final w = {w},  b = {b}")

# Then I try to plot the cost surface:

w_data = np.linspace(2, 6, 100) 
b_data = np.linspace(0, 1, 100)
w_grid, b_grid = np.meshgrid(w_data, b_data)
j_data = np.zeros_like(w_grid)

for i in range(w_grid.shape[0]):
    for j in range(w_grid.shape[1]):
        j_data[i, j] = calc_cost_function(w_grid[i, j], b_grid[i, j], bigdata)

fig = go.Figure(data=[go.Surface(
    x=w_grid,
    y=b_grid,
    z=j_data,
    colorscale="Plasma"
)])
fig.show()

What I get:

The surface looks like a 3D U shape (curved in w, almost flat in b) instead of a symmetric bowl. I thought the cost function for linear regression should always be a paraboloid bowl.

What I tried:

Narrowing/expanding the ranges of w_data and b_data.

Verified gradient descent finds reasonable parameters: w ≈ 3.93, b ≈ 0.41.

My question:

Is my plotting code wrong (meshgrid ranges, cost calculation), or is it expected that the cost surface looks like a stretched valley rather than a bowl when there’s only one feature X and one bias term b?

If it’s expected, how can I make the "bowl shape" more visible in 3D?